Alex GetmanMga maagang senyal sa AI
FIL

MiniMax M3

MiniMax

Inilabas Na-update
Context window
1M
1,000,000 token
Input / Output
larawan, teksto, video → teksto

Takdang panahon

  1. · Bersyon ng mga timbang · link sa ebidensiya
Presyo ng input / output · USD / 1 milyong token
ModeloMga link sa ebidensiyaPresyo ng input / output
minimax-m3
standard · teksto
opencode-go$0.30 / $1.20
  1. · Mga link sa ebidensiya · GitHub search · MiniMax · commits · rodrigueztaylor87/rod · @rodrigueztaylor87
  2. · Mga link sa ebidensiya · GitHub search · MiniMax · commits · SilverKnightKMA/paseo-plugins · @SilverKnightKMA
  3. · Mga link sa ebidensiya · GitHub search · MiniMax · commits · SilverKnightKMA/paseo-plugins · @SilverKnightKMA
    Sipi
    now carry the FULL pi reference '<pi-provider>/<model>' (magpie/minimax/MiniMax-M3),
  4. · Mga link sa ebidensiya · GitHub search · MiniMax · commits · Lore-Hex/quill-router · @jperla
  5. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · xiaoY233/Chat2API · @octo-patch
    Sipi
    feat(minimax): add MiniMax-M3 model entry
  6. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @Vincent-PRO-AI
    Sipi
    [Model][MiniMax-M3] SM120 Blackwell NVFP4 KV cache Triton reader
  7. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @Pb314314
    Sipi
    use sparse MLA (dense MLA, DeepSeek-V4 / V4.1, MiniMax-M3, non-MLA models) are not affected; the reason for each is in "Per-model check".
  8. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @cjackal
    Sipi
    - Pixtral, Cohere Compass and MiniMax-M3 also call `get_cu_seqlens_padding_logics`, so every encoder-CG model whose `cu_seqlens` comes from `maybe_recompute_cu_seqlens` takes its padding from one place. Without
  9. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @MrGeva
    Sipi
    [Perf][MiniMax-M3] Two-shot fused MoE finalize + all-reduce + norm + MXFP8 up to 512 tokens
  10. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @MrGeva
    Sipi
    [Perf][MiniMax-M3] Raise the fused MoE finalize + all-reduce cap from 32 to 128 tokens
  11. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @afriedri
    Sipi
    short prompts at high concurrency is higher on MiniMax-M3 (TP8: up to +21% at 1024 tokens, conc 64; TP4: +14% at 1024
  12. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · vllm-project/vllm · @wolfcomos
    Sipi
    Add `minimax_m3_msa_decode_fp16_dot`, an opt-in optimization for MiniMax-M3 Triton sparse decode on SM100-family GPUs. It uses FP16 dot operands with FP32 accumulation and selects splits targeting
  13. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · viztor/dsh-opencode-patch · @viztor
    Sipi
    | `opencode-go/minimax-m3` | info | `[KNOWN_DEFECT]` toggle-only model offered Minimal…Max | Off / On (D1) |
  14. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · thequanhcm/quanpt1 · @thequanhcm
  15. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · tenstorrent/tt-metal · @mdragulaTT
  16. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · tenstorrent/tt-metal · @mdragulaTT
    Sipi
    The MiniMax-M3 vision encoder on one Blackhole chip. It turns one image, as the media module's `prepare` resizes it, into its vision tokens: bf16 `[n_tokens,
  17. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · sgl-project/sglang · @alphabetc1
  18. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · openJiuwen-ai/jiuwenswarm · @octo-patch
  19. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · jundot/omlx · @begcdn
    Sipi
    On M2 Ultra (`applegpu_g14d`) the MiniMax-M3 sparse prefill path crashes once a prompt crosses 4096 tokens: `msa_prefill_supported` gates only on shape (`D=128`, `block_size=128`, `qhead_per_kv=16`), so
  20. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  21. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · inference-sim/inference-sim · @sriumcp
    Sipi
    the substrate's layer 8 and 9 price changes (MiniMax-M3 routers replicated in fp32, MXFP8 scale bytes, dsv4-nvfp4 MTP experts at MXFP4, a shared lm_head for
  22. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · inference-sim/blis-vllm · @sriumcp
    Sipi
    (`vllm/model_executor/models/config.py:43-55`) for every graph that is not DeepSeek-V4; MiniMax-M3's pool at the same memory is 72645 blocks, not 72893, because the catalog graph now holds
  23. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · inference-sim/blis-vllm · @sriumcp
    Sipi
    the figures on the H200 rows and the MiniMax-M3 points are development-split, in-sample numbers.
  24. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · goonercbot/hermes-agent · @goonercbot
    Sipi
    - Real configured MiniMax-M3 request through corrected Hermes adapter returned exactly `RECOVERY_OK`, end_turn, original request unchanged.
  25. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ghuntley/how-to-build-a-coding-agent · @octo-patch
    Sipi
    Anthropic-compatible defaults to the standalone workshop agents, with MiniMax-M3.1-Flash as the default and MiniMax-M3 as the alternative.
  26. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · boelabs/ai-gateway · @jhisdev
    Sipi
    - `bun run test` fails: `vercel.test.ts` pins `vercel/minimax/minimax-m3` reasoning as `fixed ["high"]`; the regenerated snapshot now says `openai_effort ["none","high"]`. Tests may not be edited by
  27. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ashhart/TensorFold · @ashhart
    Sipi
    CUDA decode attention (`window_attention`, shared by Qwen3.5/3.8, Llama, MiniMax-M3, Spark 2.5, Step5 and LFM): `window_rows` merged every chunk's partial for every row, 128 chunks at
  28. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · anomalyco/models.dev · @rjuez00
    Sipi
    - `MiniMax-M3` is no longer listed and a request using that ID returns `model=MiniMax-M3.1-Flash`, so the old entry is removed rather than kept as an
  29. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ROCm/aiter · @zhiding512
    Sipi
    On gfx1250, every MiniMax-M3 sparse attention call (decode, MTP verify, prefill) runs FlyDSL `pa_decode` (`pa_fwd_asm` does not support gfx1250), and dense layers use it with
  30. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ROCm/aiter · @RElbers
  31. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ROCm/ATOM · @yhl-amd
    Sipi
    under `multi`, which is the prefill-side setup for MiniMax-M3 PD with UMBP:
  32. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · ROCm/ATOM · @rbrugaro-amd
    Sipi
    does for the target (vLLM already did this). MiniMax-M3's DSpark draft now validates this config instead of building its own, and declares `draft_skips_online_quant` so its
  33. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · Quarkgluonmixture/quarkspace · @github-actions[bot]
    Sipi
    - `intelligence` — minimax-m3 (default) intelligence: catalog 45.4, archive 18.9
  34. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · Portkey-AI/models · @polease
  35. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · Pinvou/pinvou-agent · @asto18089
  36. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · Pb314314/vllm · @Pb314314
    Sipi
    use sparse MLA (dense MLA, DeepSeek-V4 / V4.1, MiniMax-M3, non-MLA models) are not affected; the reason for each is in "Per-model check".
  37. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · NVIDIA/TensorRT-LLM · @ziyixiong-nv
    Sipi
    sampler, speculative and QSA tests; the MXFP8 and MiniMax-M3 dispatch tests; GSM8K with DSpark (TP4);
  38. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · NVIDIA-NeMo/Megatron-Bridge · @pruprakash
    Sipi
    seeing `TEDotProductAttention`; GREEN 4/4, full file 86 passed, Gemma4/MiniMax-M3/Kimi-K2.5-VL/GPT-OSS unit tests 100 passed. E2E: gpt-oss-20b card inference (TP2 PP2 EP2, `--attention-backend local`) on this branch
  39. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · EvolvingLMMs-Lab/lmms-eval · @sahirema
    Sipi
    sample by frame rate ignore it, e.g. vLLM's MiniMax-M3 and Qwen3-VL loaders. On those, `max_frames_num` has no effect: long clips are sampled at the loader's
  40. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · overnightworks/speech · @dependabot[bot]
  41. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · maximhq/bifrost · @jimaldon
    Sipi
    model catalog datasheet already carries `tencent/deepseek-v4-pro`, `tencent/deepseek-v4-flash` and `tencent/minimax-m3`; live-only models are listed with blank pricing and can be priced via overrides.
  42. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · maximhq/bifrost · @octo-patch
    Sipi
    API. New text examples and integration defaults select `MiniMax-M3.1-Flash`, with `MiniMax-M3` as the other current choice. Default management model suggestions put those models first and
  43. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  44. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  45. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  46. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · cbc02009/container-images · @mahoro-ai
    Sipi
    `/v1/chat/completions` request to a messages-protocol Go model (e.g. `opencode_go/minimax-m3`, `qwen3.8-max`, `claude-haiku-5-5`) dies mid-stream with `TypeError: 'NoneType' object is not an iterator` → `APIConnectionError` 500 →
  47. · Mga link sa ebidensiya · GitHub search · MiniMax · pulls · MrGeva/vllm · @MrGeva
    Sipi
    320 questions, concurrency 32, same node, together with MiniMax-M3 model changes):
  48. · Pagkakalista sa katalogo · GitHub · MiniMax-AI/minimax-code · code
    Sipi
    afety margin. */ export function computeCompactionTriggerAt(input: { modelId?: string; contextWindow: number; perTurnMaxTokens: number; reserveTokens: number; safetyMarginTokens: number; }): number { if ( input.modelId === 'MiniMax-M3' && (input.contextWindow === 512_000 || input.contextWindow === 1_000_000) ) { return Math.floor(input.contextWindow * 0.9); } const perTurnMaxTokens = Number.isFinite(input.perTurnMaxTokens) && input.perTurnMaxTokens > 0 ? input.perTurnMaxTokens : 0; const reserve = Math.max(input.reserveTokens, perTurnMaxTokens + input.safetyMarginTokens); return Math.max(1, input.contextWindow - reserve); }

    Hindi pinatutunayan ng pagbanggit na ito sa code ang isang ginagamit na identifier ng modelo.

Mga espesipikasyon

Ang MiniMax M3 ay isang modelo mula sa MiniMax na may bukas na weights, na inilabas noong 2026-06-01. Ang context window ay 1,000,000 token.

Context window
1,000,000
Input
larawan · teksto · video
Output
teksto
Weights
Bukas

Mga tanong

Gaano kalaki ang context window ng MiniMax M3?

Ang context window ay 1,000,000 token.

Magagamit ba ng publiko ang weights ng MiniMax M3?

Oo. Inilalathala ng MiniMax ang weights.

Kailan inilabas ang MiniMax M3?

Noong 2026-06-01.