Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
Weights only, an estimate: leave room for the KV cache. Method
Needs an 8×80 GB node even at INT4
About 274 GB at INT4, 547 GB at FP8, 1.09 TB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.
Everything else, folded: the full tables, every source.
Use-case evidence
3 use cases
What it is good at 3 use cases
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
vLLM: not in registry · MiniMaxM1ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: not in registry · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
Hosted endpoints observed for MiniMax-M1-80k, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting. Cheapest observed input price: $0.55/M.
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
github.com2025-06-17authorityHN front pageabout the MiniMax M1 familyHN 349 pts · 75 comments1,401 words
MiniMax-M1 is the world's first open-weight, large-scale hybrid-attention reasoning model. The model contains 456 billion parameters with 45.9 billion parameters activated per token and natively supports a context length of 1 million tokens. It utilizes a novel CISPO algorithm for reinforcement learning and offers versions with 40K and 80K thinking budgets, consuming 25% of the FLOPs compared to DeepSeek R1 at 100K token generation.written, reviewed, cited
The evaluation of SWE-bench Verified excludes 14 specific test cases incompatible with the internal infrastructure.
The SWE-bench Verified score for MiniMax-M1-80K is 56.0, while DeepSeek-R1 achieves 57.6 on the same task.
The HLE benchmark result for MiniMax-M1-80K is 8.4, marked as conducted on the text-only HLE subset.
theregister.com2025-06-18about the MiniMax M1 familyHN 8 pts · 0 comments
Papers
1 paper
Papers 1
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
