Alex Getmanfrühe KI-Signale
DE

MiniMax M3

MiniMax

Veröffentlicht Aktualisiert
Kontextfenster
1M
1.000.000 Tokens
Eingabe / Ausgabe
Bild, Text, Video → Text

Zeitverlauf

  1. · Gewichtsversion · Beleglink
Preis Ein-/Ausgabe · USD / 1 Mio. Token
ModellBeleglinksPreis Ein-/Ausgabe
minimax-m3
standard · Text
opencode-go$0,30 / $1,20
  1. · Beleglinks · GitHub search · MiniMax · commits · rodrigueztaylor87/rod · @rodrigueztaylor87
  2. · Beleglinks · GitHub search · MiniMax · commits · SilverKnightKMA/paseo-plugins · @SilverKnightKMA
  3. · Beleglinks · GitHub search · MiniMax · commits · SilverKnightKMA/paseo-plugins · @SilverKnightKMA
    Auszug
    now carry the FULL pi reference '<pi-provider>/<model>' (magpie/minimax/MiniMax-M3),
  4. · Beleglinks · GitHub search · MiniMax · commits · Lore-Hex/quill-router · @jperla
  5. · Beleglinks · GitHub search · MiniMax · pulls · xiaoY233/Chat2API · @octo-patch
    Auszug
    feat(minimax): add MiniMax-M3 model entry
  6. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @Vincent-PRO-AI
    Auszug
    [Model][MiniMax-M3] SM120 Blackwell NVFP4 KV cache Triton reader
  7. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @Pb314314
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @cjackal
    Auszug
    - 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. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @MrGeva
    Auszug
    [Perf][MiniMax-M3] Two-shot fused MoE finalize + all-reduce + norm + MXFP8 up to 512 tokens
  10. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @MrGeva
    Auszug
    [Perf][MiniMax-M3] Raise the fused MoE finalize + all-reduce cap from 32 to 128 tokens
  11. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @afriedri
    Auszug
    short prompts at high concurrency is higher on MiniMax-M3 (TP8: up to +21% at 1024 tokens, conc 64; TP4: +14% at 1024
  12. · Beleglinks · GitHub search · MiniMax · pulls · vllm-project/vllm · @wolfcomos
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · viztor/dsh-opencode-patch · @viztor
    Auszug
    | `opencode-go/minimax-m3` | info | `[KNOWN_DEFECT]` toggle-only model offered Minimal…Max | Off / On (D1) |
  14. · Beleglinks · GitHub search · MiniMax · pulls · thequanhcm/quanpt1 · @thequanhcm
  15. · Beleglinks · GitHub search · MiniMax · pulls · tenstorrent/tt-metal · @mdragulaTT
  16. · Beleglinks · GitHub search · MiniMax · pulls · tenstorrent/tt-metal · @mdragulaTT
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · sgl-project/sglang · @alphabetc1
  18. · Beleglinks · GitHub search · MiniMax · pulls · openJiuwen-ai/jiuwenswarm · @octo-patch
  19. · Beleglinks · GitHub search · MiniMax · pulls · jundot/omlx · @begcdn
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  21. · Beleglinks · GitHub search · MiniMax · pulls · inference-sim/inference-sim · @sriumcp
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · inference-sim/blis-vllm · @sriumcp
    Auszug
    (`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. · Beleglinks · GitHub search · MiniMax · pulls · inference-sim/blis-vllm · @sriumcp
    Auszug
    the figures on the H200 rows and the MiniMax-M3 points are development-split, in-sample numbers.
  24. · Beleglinks · GitHub search · MiniMax · pulls · goonercbot/hermes-agent · @goonercbot
    Auszug
    - Real configured MiniMax-M3 request through corrected Hermes adapter returned exactly `RECOVERY_OK`, end_turn, original request unchanged.
  25. · Beleglinks · GitHub search · MiniMax · pulls · ghuntley/how-to-build-a-coding-agent · @octo-patch
    Auszug
    Anthropic-compatible defaults to the standalone workshop agents, with MiniMax-M3.1-Flash as the default and MiniMax-M3 as the alternative.
  26. · Beleglinks · GitHub search · MiniMax · pulls · boelabs/ai-gateway · @jhisdev
    Auszug
    - `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. · Beleglinks · GitHub search · MiniMax · pulls · ashhart/TensorFold · @ashhart
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · anomalyco/models.dev · @rjuez00
    Auszug
    - `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. · Beleglinks · GitHub search · MiniMax · pulls · ROCm/aiter · @zhiding512
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · ROCm/aiter · @RElbers
  31. · Beleglinks · GitHub search · MiniMax · pulls · ROCm/ATOM · @yhl-amd
    Auszug
    under `multi`, which is the prefill-side setup for MiniMax-M3 PD with UMBP:
  32. · Beleglinks · GitHub search · MiniMax · pulls · ROCm/ATOM · @rbrugaro-amd
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · Quarkgluonmixture/quarkspace · @github-actions[bot]
    Auszug
    - `intelligence` — minimax-m3 (default) intelligence: catalog 45.4, archive 18.9
  34. · Beleglinks · GitHub search · MiniMax · pulls · Portkey-AI/models · @polease
  35. · Beleglinks · GitHub search · MiniMax · pulls · Pinvou/pinvou-agent · @asto18089
  36. · Beleglinks · GitHub search · MiniMax · pulls · Pb314314/vllm · @Pb314314
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · NVIDIA/TensorRT-LLM · @ziyixiong-nv
    Auszug
    sampler, speculative and QSA tests; the MXFP8 and MiniMax-M3 dispatch tests; GSM8K with DSpark (TP4);
  38. · Beleglinks · GitHub search · MiniMax · pulls · NVIDIA-NeMo/Megatron-Bridge · @pruprakash
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · EvolvingLMMs-Lab/lmms-eval · @sahirema
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · overnightworks/speech · @dependabot[bot]
  41. · Beleglinks · GitHub search · MiniMax · pulls · maximhq/bifrost · @jimaldon
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · maximhq/bifrost · @octo-patch
    Auszug
    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. · Beleglinks · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  44. · Beleglinks · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  45. · Beleglinks · GitHub search · MiniMax · pulls · iv-org/invidious · @Fijxu
  46. · Beleglinks · GitHub search · MiniMax · pulls · cbc02009/container-images · @mahoro-ai
    Auszug
    `/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. · Beleglinks · GitHub search · MiniMax · pulls · MrGeva/vllm · @MrGeva
    Auszug
    320 questions, concurrency 32, same node, together with MiniMax-M3 model changes):
  48. · Katalogeintrag · GitHub · MiniMax-AI/minimax-code · code
    Auszug
    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); }

    Diese Codereferenz bestätigt keinen tatsächlich verwendeten Modellbezeichner.

Technische Daten

MiniMax M3 ist ein Modell von MiniMax mit offenen Modellgewichten, veröffentlicht am 2026-06-01. Das Kontextfenster umfasst 1.000.000 Tokens.

Kontextfenster
1.000.000
Eingabe
Bild · Text · Video
Ausgabe
Text
Modellgewichte
Offen

Fragen

Wie groß ist das Kontextfenster von MiniMax M3?

Das Kontextfenster umfasst 1.000.000 Tokens.

Sind die Modellgewichte von MiniMax M3 offen?

Ja. MiniMax veröffentlicht die Modellgewichte.

Wann wurde MiniMax M3 veröffentlicht?

Am 2026-06-01.