MiniMax launches M3.1 Flash at $0.15 per million input tokens, with open weights planned
The Chinese lab says the model is faster and 70% cheaper on agent tasks than its predecessor. Independent tests from Artificial Analysis show strong scores and a high hallucination rate on one test.
Illustration · not a photoHow this story was checked
- ResearchedWith AI assistance · 2 sources
- AI source checkEach sentence matched to its source
- Editor approvalMayank Sahu · 20 Oct, 13:25
- Published20 Oct, 13:25
Verify this story
- 2 sources, 1 official · 1 with an archived copy · see sources
- Every fact has a label and a source number · fact ledger
- What we searched, read and could not verify · research note
- Machine-readable: facts as JSON · text as Markdown
How this was made: researched, drafted and source-checked by AI (Claude, made by Anthropic), then published under the editor’s standing approval. A person did not read it line by line before it went live. AI policy
In short
MiniMax released M3.1 Flash on 19 October 2026, priced at $0.15 per million input tokens and $0.60 per million output tokens through its API. The company says agent tasks cost 70% less than on its previous model and that open weights will follow.
The launch
MiniMax says M3.1 Flash is available now through the MiniMax API, its M Plan subscription and MiniMax Code.1 The model was tested in public under the codename “Space Bunny” before launch.2
Price
- Input: $0.15 per million tokens.12
- Output: $0.60 per million tokens.12
- Cached input: $0.015 per million tokens.12
MiniMax says agent tasks cost 70% less than on its predecessor and that output speed exceeds 200 tokens per second, with median speeds above 170.1
Benchmarks
MiniMax reports its own results, including 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1 and 61.7 on OSWorld 2.0.1
Independent tests by Artificial Analysis, reported by VentureBeat, give M3.1 Flash 42 on its Intelligence Index, up from 29 for M3, at about $0.22 per task.2 On Artificial Analysis’s knowledge test, AA-Omniscience, it answered 56% correctly, with a 93% hallucination rate. That rate measures how often its non-correct answers were confidently wrong rather than declined, not the share of all answers.2 For fact-heavy uses, that means checking outputs remains important.
Open weights
MiniMax says weights will be released later.1 VentureBeat reports the planned date as 23 October; licence terms have not been announced.2 VentureBeat also gives a 1 million token context window.2 We will update when weights appear.
Why it matters
Prices for capable, fast models keep falling. M3.1 Flash adds pressure at the low end, where many agent workloads run. See our API price tracker for comparisons.
Fact ledger
How labels workTracked from this story
All trackers| Model | Tier | Input | Output | From | Label |
|---|---|---|---|---|---|
| minimax-m3.1-flash | Standard | $0.15 | $0.60 | 19 Oct 2026 | Confirmed |
Questions readers ask
How much does MiniMax M3.1 Flash cost?
$0.15 per million input tokens, $0.60 per million output tokens and $0.015 per million cached input tokens.
Is MiniMax M3.1 Flash open source?
Not yet. MiniMax says weights will be released later; VentureBeat reports a planned date of 23 October 2026.
What does the 93% hallucination rate mean?
On Artificial Analysis’s AA-Omniscience test, 93% of its non-correct answers were wrong rather than declined. It is not the share of all answers.