Skip to content
Menu

StrataLLM

$STRATA

AI · Utility

Contract 0x7637e30548afb698e41534e6c86fcd5209466fef

HEY holds 170 other Robinhood Chain tokens that use $STRATA. This page is only the contract above. See them

Is this your project? Claim and verify →

Run a 125B AI model on your own gaming PC. Benchmarks, real-time replays and setup guides for the open-source Strata engine. Community project.

What HEY knows

Building
Shipping consistentlyShipping: 16 meaningful events in 30 days; latest ship yesterday. Evidence Source

Build Momentum 70How it is measured

Latest ship
Strata 0.1.45 · · GitHub release

Why it matters: One of a run of ships in the 30 days to it: building continued, not a one-off.

FollowCreate alertTerminal previewView latest ship
Share

Follow puts StrataLLM on your private watchlist; its changes then lead “Since you were last here” when you come back. Following needs an account; browsing never does.

No account? Email me StrataLLM’s changes

No account and no password. HEY stores the address and what you asked for, sends one confirmation, and nothing else until you confirm. Every email has a one-click unsubscribe that deletes the address.

Or read its RSS feed.

Product usage
UnknownThe project has a contract, but HEY does not count its calls yet: a token joins the watch once it takes calls in a thirty-day window, and any contract the project’s own sources name is watched. Unknown, not zero.
Contracts
Fact1 contract on record for code changes; no implementation or interface change recorded.
Token market · context
Market data insufficient$13K fully diluted valuationToken verified
Not yet known
Unknown8 gaps: API description (not enough yet), Official site changes (not enough yet), Contract creation (not enough yet), Verified source (not enough yet), Contract interface (not enough yet), Current market (not enough yet), Market history (not enough yet), Supply distribution (not enough yet). HEY has not read its website’s published files yet.

HEY can answer 11 of the 19 research questions that apply · Research coverage

Compare with…The same summary, with every line’s evidence: snapshot API
What HEY checked — why it says this
Shipping.
“Shipping” by the activity rule: 16 meaningful events in the last 30 days, the newest 2026-10-17.
SourceRepositories, changelogs and feeds HEY reads for this project, and the ship events recorded from them.ReadRecordsStrata 0.1.45Code changes this week (since Mon 12 Oct) · 75 commitsStrata 0.1.44.1+2 moreWhat HEY derivedHEY’s activity rule reads the counted events as “Shipping”.Not concludedIt says what HEY can read of the building, not how good the project is, who is behind it, or what its token will do. Quiet or unknown is not "abandoned".
Verified Builder.
The quality gate found verified building evidence from the project's own sources.
Source4 registered sources, 0 of them context only.ReadRecordsStrata 0.1.45Code changes this week (since Mon 12 Oct) · 75 commitsStrata 0.1.44.1+6 moreWhat HEY derivedThe quality gate’s stored decision: “Verified Builder”.Not concludedA verified builder is one whose shipping HEY confirmed from public evidence. It does not say who runs the project — that is the ownership claim, a separate check.
Token verified.
Token verified: the project's own site, a deploy record in its own repository, or an on-chain signature names this contract.
SourceThe project's own published sources and HEY's check of the contract.ReadWhat HEY derivedToken verified: the project’s own site names it.Not concludedIt says whether the project itself names this contract. It is not a safety check, and a token is not the project.
Market data insufficient.
Market data insufficient: HEY holds no market reading for the token.
Source2 readings from HEY's own market tables.ReadWhat HEY derivedHEY’s market rule reads this token’s market as “Market data insufficient”.Not concludedMarket status is context only. It never changes activity status or Build Momentum, and it is not a view on the price.
Market figures are context only and never change activity status. Methodology
LinksWebsiteGitHubXdefined.fiRHToolsSources

Latest ship

Checked

Strata 0.1.45

GitHub releaseSource verified

0.1.45: Qwen3.6-35B-A3B reads pictures, serves several requests at once and runs up to 40% faster; switches for multi-user servers; Mac and setup fixes. Answers of the Qwen3.8-Flash-Next models are the same as in 0.1.44.1 (byte-identical in our identity tests on every GPU… Show all
  • 0.1.45: Qwen3.6-35B-A3B reads pictures, serves several requests at once and runs up to 40% faster; switches for multi-user servers; Mac and setup fixes. Answers of the Qwen3.8-Flash-Next models are the same as in 0.1.44.1 (byte-identical in our identity tests on every GPU machine we tested: CUDA sm_60 / sm_86, HIP gfx1201, Intel A750, Mac M4). Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.45. Qwen3.6-35B-A3B
  • Pictures. Setup asks "Do you want images?" for Qwen3.6, downloads the pinned mmproj (0.9 GB) and writes --vision. It is the same…

Record ship:ef6744d6… · HEY counts it as building; it does not judge the release’s quality or what it means for the token.

View evidence ↗Post this ship on X

Builder activity

Bars: meaningful, source-backed ships per week (count). Weeks run Monday to Sunday, UTC, labelled by their Monday. Commits are aggregated, a code week counted once. Hatched: 9 weeks before HEY began reading this project’s sources — not yet read, not zero.

Questions HEY can answer about StrataLLM — as of 2026-10-18

Is StrataLLM still being built?
DerivedShipping: 16 meaningful events in 30 days; latest ship yesterday. Evidence
What did StrataLLM ship last?
FactStrata 0.1.45 — GitHub release, yesterday. Evidence
Is 0x7637e30548aFb698E41534e6C86FCd5209466Fef StrataLLM’s own contract?
FactToken verified: The project’s own site names this contract. Details
What does HEY not know about StrataLLM?
Unknown8 gaps: API description (not enough yet), Official site changes (not enough yet), Contract creation (not enough yet), +5 more. HEY has not read its website’s published files yet. Details

Build timeline

  1. Code changes this week (since Mon 12 Oct) · 75 commits

    Code activitySource linked

    What changed, in the repository’s own words:

    5 active days in niko1221/strata.

    This week HEY looked inside 5 of 75 commits to see what changed: 3 changed code, 1 documentation or maintenance only, 1 could not be classified. In a burst week, enough of a sample has to change code for the week to count.

    View evidence

  2. Strata 0.1.44.1

    GitHub releaseSource verified

    A hotfix for 0.1.44: PCs with an older CPU start again, CUDA 12.0 / 12.1 toolkits can build the engine, and Apple M2 Macs give correct answers. Answers are the same as 0.1.44 (byte-identical in our identity tests on every GPU machine we tested). Update with UPDATE.bat (Linux… Show all
    • A hotfix for 0.1.44: PCs with an older CPU start again, CUDA 12.0 / 12.1 toolkits can build the engine, and Apple M2 Macs give correct answers. Answers are the same as 0.1.44 (byte-identical in our identity tests on every GPU machine we tested). Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.44.1. Fixes
    • CPUs without AVX2 (Xeon X5670 and older, Sandy/Ivy Bridge): the engine stopped at start with nothing in the log (#2324, thanks almirjakubaitis). 0.1.44's NVFP4 CPU code built a small lookup table when the program started, in two files that are…
    View evidence

  3. Strata 0.1.44

    GitHub releaseSource verified

    Update: the issues found right after this release (Macs other than M4/M5, building with CUDA 12.1, Pascal cards on CPUs without AVX) are fixed in 0.1.44.1. Please use that one. 0.1.44: Unsloth's 4- to 8-bit files and NVFP4 packs in setup, Qwen3.6-35B-A3B, faster multi-GPU on AMD… Show all

    Update: the issues found right after this release (Macs other than M4/M5, building with CUDA 12.1, Pascal cards on CPUs without AVX) are fixed in 0.1.44.1. Please use that one. 0.1.44: Unsloth's 4- to 8-bit files and NVFP4 packs in setup, Qwen3.6-35B-A3B, faster multi-GPU on AMD and Arc A-series cards, and a long list of fixes. Answers on the default models are the same as 0.1.43.1 (byte-identical in our identity tests on every machine we tested), with the exceptions listed under "Defaults that changed" (RDNA4, and the…

    View evidence

  4. Strata v0.1.43.1

    GitHub releaseSource verified

    A hotfix for 0.1.43: Linux NVIDIA decode is fast again on desktops (the adaptive overlap is opt-in again), long prompts no longer stall on Windows and SATA SSDs, two-card batch serving is about 35% faster again, RX 6000 prompts are fast again, and fixes for AMD multi-slot… Show all

    A hotfix for 0.1.43: Linux NVIDIA decode is fast again on desktops (the adaptive overlap is opt-in again), long prompts no longer stall on Windows and SATA SSDs, two-card batch serving is about 35% faster again, RX 6000 prompts are fast again, and fixes for AMD multi-slot serving, Intel cards, hangs and the server. Answers are the same as 0.1.43 (byte-identical in our identity tests on every machine we tested); only a user message that quotes </tool_call> is now read as text (#2026). Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.43.1. Fixes…

    View evidence

  5. Strata v0.1.43

    GitHub releaseSource verified

    Faster decode and prompts on NVIDIA, AMD and Intel, plus the fixes since 0.1.42. Decode is up to 38% faster on a 16 GB Radeon, 5-11% on RTX 30-series and Tesla P100 under Linux, up to 19% on four Radeons, and 14-21% on an Arc Pro B70, whose prompts read about twice as fast.… Show all

    Faster decode and prompts on NVIDIA, AMD and Intel, plus the fixes since 0.1.42. Decode is up to 38% faster on a 16 GB Radeon, 5-11% on RTX 30-series and Tesla P100 under Linux, up to 19% on four Radeons, and 14-21% on an Arc Pro B70, whose prompts read about twice as fast. Windows streaming from GGUF files decodes 1.35-1.85x faster. Answers are the same as 0.1.42 (byte-identical on every machine we tested) except where a default below says otherwise. Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.43. Speed All numbers are our own interleaved A/Bs…

    View evidence

  6. Strata v0.1.42

    GitHub releaseSource verified

    About 13-16% faster decode on one NVIDIA GPU or a layer split, up to 30% more on a single card whose CPU outruns its PCIe link, a long list of fixes for multi-GPU splits, Windows AMD, Intel Arc and tight-RAM machines, and new opt-in speedups for AMD and multi-GPU cards. Two new… Show all

    About 13-16% faster decode on one NVIDIA GPU or a layer split, up to 30% more on a single card whose CPU outruns its PCIe link, a long list of fixes for multi-GPU splits, Windows AMD, Intel Arc and tight-RAM machines, and new opt-in speedups for AMD and multi-GPU cards. Two new NVIDIA defaults change answers a little (STRATA_ROUTE_TAIL_SKIP=7 and the measured PCIe share; each has a setting that turns it off). Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.42. The five fixes planned as 0.1.41.1 were never released and are part of this version.…

    View evidence

  7. Strata v0.1.41

    GitHub releaseSource verified

    Faster on more than one GPU, faster short prompts on NVIDIA, about twice as fast prompts on Windows with little RAM, and a long list of fixes. Two defaults change: --batch on a layer split now runs one pipeline group per GPU, and short prompt chunks on one NVIDIA GPU get help… Show all

    Faster on more than one GPU, faster short prompts on NVIDIA, about twice as fast prompts on Windows with little RAM, and a long list of fixes. Two defaults change: --batch on a layer split now runs one pipeline group per GPU, and short prompt chunks on one NVIDIA GPU get help from the CPU (this one changes the last bits of an answer; one setting turns it off). Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.41. This release includes the Pascal decode fix from 0.1.40.4. Speed Against 0.1.40.4, default settings, interleaved pairs, medians (prompt =…

    View evidence

  8. Strata v0.1.40.4

    GitHub releaseSource verified

    Hotfix for 0.1.40.3: decode on Pascal cards (Tesla P40, GTX 1080, GTX 1070) is back to the 0.1.40 speed. Default answers are unchanged. Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.40.4. Pascal decode (#1469, #1470) · Decode on sm_61 had halved… Show all
    • Hotfix for 0.1.40.3: decode on Pascal cards (Tesla P40, GTX 1080, GTX 1070) is back to the 0.1.40 speed. Default answers are unchanged. Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.40.4. Pascal decode (#1469, #1470)
    • Decode on sm_61 had halved since 0.1.40.2: the small-batch matrix kernels lost their __restrict__ hints and gained a prefetch that Pascal cannot overlap, so every decode step ran roughly twice as long. The reporters' own version of this fix took a P40 from 17.7 back to 36.7 tok/s; we have no sm_61 card, so a confirmation on…
    View evidence

  9. Strata v0.1.40.3

    GitHub releaseSource verified

    Hotfix for 0.1.40.2: Intel Arc install and A-series fixes, Windows AMD speed and runtime fixes, Docker config links, and a few smaller fixes. Default answers are unchanged. Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.40.3. Intel Arc · Install… Show all
    • Hotfix for 0.1.40.2: Intel Arc install and A-series fixes, Windows AMD speed and runtime fixes, Docker config links, and a few smaller fixes. Default answers are unchanged. Update with UPDATE.bat (Linux: ./update.sh); setup replaces the engine with 0.1.40.3. Intel Arc
    • Install: python3 sycl/setup_intel.py, as the 0.1.40.2 notes said, fails on Ubuntu's managed Python. The way in is: sudo apt install docker.io, add yourself to the docker group and log in again, then ./setup.sh --backend sycl. docs/INTEL.md now walks through it, including the build (it was missing) and…
    View evidence

  10. Strata v0.1.40.2

    GitHub releaseSource verified

    Faster and fixed across the board: official Intel Arc support (Arc Pro B70 and A750 tested), bug fixes and speed-ups merged from the open issues and PRs, and big gains on multi-GPU, low-RAM Linux and long-document use. Default answers are byte-identical to 0.1.40 (checked on… Show all

    Faster and fixed across the board: official Intel Arc support (Arc Pro B70 and A750 tested), bug fixes and speed-ups merged from the open issues and PRs, and big gains on multi-GPU, low-RAM Linux and long-document use. Default answers are byte-identical to 0.1.40 (checked on Q2_0, IQ3_XXS, Coder and IQ3_S). See Updating below if your copy is older than 0.1.40.1. Speed Measured against 0.1.40 in interleaved pairs (medians, 10 pairs unless noted): | Machine | Model | Decode (story / code) | Prompt 4K / 20K | |---|---|---|---| | RTX 5070 12 GB, Windows | IQ3_XXS | 48.5 -> 50.3…

    View evidence

  11. Strata v0.1.40.1

    GitHub releaseSource verified

    Hotfix for 0.1.40: a tool call that is only quoted (in the thinking or in a code block) no longer runs, and requests waiting during an engine restart no longer hang. Python server fixes only. The engine is the same as in 0.1.40, so there is nothing new to download. See Updating… Show all

    Hotfix for 0.1.40: a tool call that is only quoted (in the thinking or in a code block) no longer runs, and requests waiting during an engine restart no longer hang. Python server fixes only. The engine is the same as in 0.1.40, so there is nothing new to download. See Updating below: this one time it needs one extra command. Tool calls written inside the thinking (#804, #1058): 0.1.40 turned a <tool_call> the model wrote inside its thinking into a real call whenever the tool was one of the request's. That rescue is meant for a model that forgets and then writes its…

    View evidence

  12. Strata v0.1.40

    GitHub releaseSource verified

    Strix Halo (gfx1151) support, crash, NaN and security fixes, better multi-GPU, batching and resident modes, Q4_0 / Q4_1 and more PLE table formats, AMD fixes, a much faster Linux cold start, and a long list of opt-ins. Strix Halo / Ryzen AI Max (gfx1151, experimental): the HIP… Show all

    Strix Halo (gfx1151) support, crash, NaN and security fixes, better multi-GPU, batching and resident modes, Q4_0 / Q4_1 and more PLE table formats, AMD fixes, a much faster Linux cold start, and a long list of opt-ins. Strix Halo / Ryzen AI Max (gfx1151, experimental): the HIP engine now runs on the Radeon 8060S / 8050S, and setup recognizes the chip (PCI 1002:1586), sizes memory for it as one shared pool, recommends UD-IQ4_XS, and on Linux picks HIP by itself. The engine has RDNA3.5 matrix-core kernels for the prompt, a hipBLASLt table for ROCm 7.14.1, free-memory accounting for…

    View evidence

  13. Strata v0.1.39

    GitHub releaseSource verified

    Faster decode and long prompts, several requests at once, older and other hardware as experimental opt-ins, the OpenAI Responses API for Codex, a fix for slower prompts with a RAM budget, and a batch of fixes from your reports. Faster (the same answers where nothing says… Show all
    • Faster decode and long prompts, several requests at once, older and other hardware as experimental opt-ins, the OpenAI Responses API for Codex, a fix for slower prompts with a RAM budget, and a batch of fixes from your reports. Faster (the same answers where nothing says otherwise):
    • Decode (#646): the verify pass runs with fewer launches and host round trips (sub-warp expert packing, staged inputs, the PLE and MTP steps batched). Measured here on one RTX 5070 against 0.1.38, 10 interleaved pairs, medians: Q2_0 +6% (story 68.4 -> 72.7 tok/s, code 75.8 -> 80.0), IQ3_XXS +6%…
    View evidence

  14. Strata v0.1.38

    GitHub releaseSource verified

    Faster prompts, --kv q4_0 prompts on tensor cores, a 6 GB card that starts, and a large batch of community PRs. Security: the server without an API key (DNS rebinding, cross-site requests). Without api_key, the server now answers only requests addressed to a name it knows… Show all

    Faster prompts, --kv q4_0 prompts on tensor cores, a 6 GB card that starts, and a large batch of community PRs. Security: the server without an API key (DNS rebinding, cross-site requests). Without api_key, the server now answers only requests addressed to a name it knows (localhost, an IP address, the address it listens on and, when it listens beyond this PC, this PC's name), and a browser page from another site can no longer send it requests (403). /unload and /load take only JSON: curl -X POST -H "Content-Type: application/json" localhost:8080/unload. Clients without…

    View evidence

  15. Strata v0.1.37

    GitHub releaseSource verified

    A frozen engine no longer hangs the server, Windows AMD counts the desktop's VRAM, and a batch of setup fixes from your reports. The server notices a silent engine (#481): if the engine prints nothing during a request for engine_silence_s seconds (default 300, longer while a… Show all

    A frozen engine no longer hangs the server, Windows AMD counts the desktop's VRAM, and a batch of setup fixes from your reports. The server notices a silent engine (#481): if the engine prints nothing during a request for engine_silence_s seconds (default 300, longer while a long prompt is being read; 0 turns it off), or never answers a stop, the server ends the request with an error and the next request restarts the engine. Before, both sides could wait for each other forever. The cause of the reported freeze is still being traced with the reporter. **AMD on Windows (#497, PRs…

    View evidence

  16. Strata v0.1.36

    GitHub releaseSource verified

    Faster: Q2_0 reads prompts 16-22% faster and writes up to 18% faster at long contexts. Plus one-click updates (UPDATE.bat), an opt-in memory of which experts you use, and clearer logs. Speed (#136), measured on an RTX 5070 12 GB, 0.1.35 -> 0.1.36: | Q2_0 | Reads your prompt |… Show all
    • Faster: Q2_0 reads prompts 16-22% faster and writes up to 18% faster at long contexts. Plus one-click updates (UPDATE.bat), an opt-in memory of which experts you use, and clearer logs. Speed (#136), measured on an RTX 5070 12 GB, 0.1.35 -> 0.1.36: | Q2_0 | Reads your prompt | Writes answers | | --- | ---: | ---: | | 4K prompt | 1,294 -> 1,570 tokens/s (+21%) | 89 -> 93.5 (+5%) | | 32K prompt | 2,170 -> 2,653 (+22%) | 80 -> 81 | | 128K prompt | 2,123 -> 2,468 (+16%) | 64.5 -> 76.4 (+18%) |
    • New prompt kernels for Q2_0 (RTX 30 and newer): the experts are…
    View evidence

  17. Strata v0.1.35

    GitHub releaseSource verified

    Windows AMD no longer crashes on the first prompt, the low-RAM mode works on 32 GB Windows PCs, and a batch of fixes from your reports. AMD on Windows (#468, #461): the engine now uses the HIP runtime it ships with. Windows looks for a program's DLLs in its own folder, then in… Show all

    Windows AMD no longer crashes on the first prompt, the low-RAM mode works on 32 GB Windows PCs, and a batch of fixes from your reports. AMD on Windows (#468, #461): the engine now uses the HIP runtime it ships with. Windows looks for a program's DLLs in its own folder, then in System32, and only then on PATH, where the engine's rocm\bin is. AMD drivers that install their own amdhip64_7.dll in System32 won, and the bundled libraries crashed on the first prompt (an access violation on a W7900, hipErrorInvalidDeviceFunction on a 7900 XTX). Setup now puts the bundled runtime next…

    View evidence

  18. Strata v0.1.34

    GitHub releaseSource verified

    AMD cards on Windows (new), an MCP server so your AI assistant can set Strata up, a shorter README · and a cancelled request now frees the engine within a second. AMD on Windows (new, #247 #325): an AMD card on Windows is set up like an NVIDIA one: double-click START-HERE.bat.… Show all
    • AMD cards on Windows (new), an MCP server so your AI assistant can set Strata up, a shorter README
    • and a cancelled request now frees the engine within a second. AMD on Windows (new, #247 #325): an AMD card on Windows is set up like an NVIDIA one: double-click START-HERE.bat. On a PC with no NVIDIA card Strata can use, setup picks the AMD card by itself and downloads the ready-made AMD engine (strata-windows-x64-hip.zip, below). It carries the ROCm libraries it needs, so the AMD driver (AMD Software: Adrenalin Edition) is the only thing to install. Cards: RX 7900 XT / XTX, RX…
    View evidence

  19. Strata v0.1.33

    GitHub releaseSource verified

    The image encoder runs on every CPU again, and setup recommends instead of forcing: a longer context, more GPUs or a bigger RAM budget you choose is kept, with a note about the risk. Images (#411, #412): 0.1.32's ready-made strata-vision.exe was built with AVX-512 and OpenMP and… Show all

    The image encoder runs on every CPU again, and setup recommends instead of forcing: a longer context, more GPUs or a bigger RAM budget you choose is kept, with a note about the risk. Images (#411, #412): 0.1.32's ready-made strata-vision.exe was built with AVX-512 and OpenMP and stopped at start on CPUs without AVX-512 (most AMD Ryzen before Zen 4, many Intel desktop CPUs): "illegal instruction" or "the vision encoder did not start". 0.1.33's is built portable again (AVX2, no OpenMP), like 0.1.31's, and keeps 0.1.32's RTX 20 code. Running START-HERE.bat once installs it. **Setup…

    View evidence

  20. Strata v0.1.32

    GitHub releaseSource verified

    Multi-GPU short prompts fixed, AMD decode +12%, Unsloth's Q4 model in setup with 2.8x faster prompts, and 30+ community changes · with the same answers. Multi-GPU: short prompts faster again (#340, riverhh76). Since 0.1.30 a layer split lent each card's expert cache to its… Show all
    • Multi-GPU short prompts fixed, AMD decode +12%, Unsloth's Q4 model in setup with 2.8x faster prompts, and 30+ community changes
    • with the same answers. Multi-GPU: short prompts faster again (#340, riverhh76). Since 0.1.30 a layer split lent each card's expert cache to its prompt path and refilled it after every request, one card after another; 0.1.31's bigger streaming ring made the loans larger. Short prompts lost up to a third of their speed (2x RTX A5000: -35%). 0.1.32 refills all cards at once, uses a smaller ring on a split, and lets a card with free VRAM keep its own prompt…
    View evidence

  21. Strata v0.1.31

    GitHub releaseSource verified

    Unsloth's Q4 model runs on a 64 GB PC (experimental), AMD decode up to 15% faster, models load twice as fast on Windows, and a batch of fixes. Experimental: Unsloth UD-Q4_K_XL on a normal PC. Strata can now run Unsloth's 4-bit Qwen3.8-Flash-Next (UD-Q4_K_XL, 111 GB in 4 files… Show all
    • Unsloth's Q4 model runs on a 64 GB PC (experimental), AMD decode up to 15% faster, models load twice as fast on Windows, and a batch of fixes. Experimental: Unsloth UD-Q4_K_XL on a normal PC. Strata can now run Unsloth's 4-bit Qwen3.8-Flash-Next (UD-Q4_K_XL, 111 GB in 4 files, 72 GiB of experts) on a PC whose RAM is smaller than the model:
    • The model's four files are read as they are: a missing file is an error, a layer split across files is fine.
    • Its expert formats (Q4_K, Q5_K, Q5_1, Q8_0) run natively on the GPU and the CPU; the GPU dequantizers match llama.cpp's bit for bit…
    View evidence

  22. Strata v0.1.30

    GitHub releaseSource verified

    Short prompts up to 28% faster (the same answers), multi-GPU and AMD RDNA4 improvements, a server that can give the GPU back, and several opt-in features. Faster short prompts. Prompts of 1K to 4K tokens read all experts from the stream from 1,024-token chunks (0.1.29: 2,048)… Show all
    • Short prompts up to 28% faster (the same answers), multi-GPU and AMD RDNA4 improvements, a server that can give the GPU back, and several opt-in features. Faster short prompts. Prompts of 1K to 4K tokens read all experts from the stream from 1,024-token chunks (0.1.29: 2,048): +17-28% prompt speed on this RTX 5070, with the same output (suggested in the issues). STRATA_PREFILL_STREAM_MIN overrides the threshold. Multi-GPU (#216, @gopinath87607). In a layer split:
    • Each card now keeps only its own layers' session state, which saved 0.93 GiB on a later card in the author's 4-GPU…
    View evidence

  23. Strata v0.1.29

    GitHub releaseSource verified

    Sampled answers up to 40% faster (the same text), faster prompt kernels, correctness fixes, and several contributions. Sampled answers (#197, @gputier). With a temperature above 0 (most chat apps), each token's top-k selection now runs split across the whole GPU, and the rest of… Show all
    • Sampled answers up to 40% faster (the same text), faster prompt kernels, correctness fixes, and several contributions. Sampled answers (#197, @gputier). With a temperature above 0 (most chat apps), each token's top-k selection now runs split across the whole GPU, and the rest of the sampling runs on one warp. It picks exactly the same tokens: with the same seed, the output is identical to 0.1.28 (checked on 12 runs across 3 sampling settings, and by sampler_parity against the old kernel). On an RTX 5070:
    • Q2_0: +4% with top-k 20, +15-20% with top-k 40, +38-42% with top-k 64.
    • The…
    View evidence

  24. Strata v0.1.28

    GitHub releaseSource verified

    Fixes: VRAM left free again after 0.1.27's larger draft head, a cancelled request no longer kills the next one, and the API key protects every endpoint. VRAM (#199, #201, #217). The expert cache takes the free VRAM minus a reserve (700 MiB) that keeps the GPU from paging. The… Show all

    Fixes: VRAM left free again after 0.1.27's larger draft head, a cancelled request no longer kills the next one, and the API key protects every endpoint. VRAM (#199, #201, #217). The expert cache takes the free VRAM minus a reserve (700 MiB) that keeps the GPU from paging. The draft layer's head is allocated after the cache, and 0.1.27's larger head (with Chinese, Japanese and Korean) came out of that reserve: on a 16 GB card with images, 156 MiB were left, and requests could stall. The cache now reserves the draft head as well, so the reserve is kept. Auto-sized caches hold a few…

    View evidence

Research notes

Written by HEY researchers and reviewed before publishing. Commentary with sources — not the project's own claims, and not evidence of shipping.

Know something about StrataLLM? Sign in with GitHub to submit a research note. Notes are reviewed before they appear here.

What changed

32 changes on HEY’s record for StrataLLM: 30 releases · 1 new project on HEY · 1 week of code pushed.

All 32 changes for StrataLLM →

The full change record holds 2 more, of kinds this list leaves out — lock records and HEY’s own research bookkeeping among them. Full record (JSON)

Research coverage

HEY can answer 11 of the 19 research questions that apply.

11 of 19 applicable dimensions measuredShow each dimension

Identity

  • IdentityMeasuredRead 1h ago.

Building

  • Building activityMeasuredRead 57m ago.
  • RepositoriesMeasured
  • ReleasesMeasuredRead 1d ago.
  • TimelineMeasured

Official site

  • Official docsMeasuredRead 1h ago.
  • API descriptionNot enough yetHEY has not read its website’s published files yet.
  • Official site changesNot enough yetHEY has no first reading of its website to compare against yet.

Code and packages

  • Repository detailsMeasuredRead 30m ago.
  • Published packagesMeasuredRead 30m ago.

Contracts

  • Contract creationNot enough yetHEY has not read the contract’s creation yet.
  • Contract activityMeasuredRead today.
  • Verified sourceNot enough yetHEY has not read the contract’s published source yet.
  • Contract interfaceNot enough yetHEY has not read the contract’s published source yet.

Market

  • Current marketNot enough yetNo market reading yet.
  • Market historyNot enough yetNo daily market history recorded yet.
  • Supply distributionNot enough yetHEY has not mapped this token’s supply yet.
  • LocksMeasuredRead 2h ago.
  • Market checksMeasured

Not applicable here: Package advisories, Protocol economics.

A count of what HEY has read, never a score: projects are not compared or ordered by it. Unknown is never zero. The same states: coverage API.

Sources

Contract
0x7637…6fef0x7637e30548afb698e41534e6c86fcd5209466fef