MiniMax M3.1 Flash vs Qwen3.8 27B
Full profiles: MiniMax M3.1 Flash (MiniMax) and Qwen3.8 27B (Alibaba): every benchmark, price and rank.
Updated · Data from 10 sources
As of 16 Nov 2026, MiniMax M3.1 Flash and Qwen3.8 27B are neck and neck on quality (54.3 vs 55.1 on the BenchLeader Index, inside the ±7.2 margin), and MiniMax M3.1 Flash is 4.2× cheaper at $0.262 vs $1.09 per 1M tokens blended. Choose MiniMax M3.1 Flash for agents & tools (55 vs 44), where it also streams ~2× faster; choose Qwen3.8 27B for fast first answers (0.7 s vs 43.3 s) and reasoning (58 vs 50) and coding (62 vs 55).
Key facts (as of 16 Nov 2026)
- MiniMax M3.1 Flash scores 54.3 on the BenchLeader Index and Qwen3.8 27B scores 55.1 (BenchLeader, 16 Nov 2026).
- MiniMax M3.1 Flash costs $0.262 per 1M tokens blended (list price, 16 Nov 2026).
- Qwen3.8 27B costs $1.09 per 1M tokens blended (list price, 16 Nov 2026).
- MiniMax M3.1 Flash streams ~2× faster than Qwen3.8 27B (Artificial Analysis, 7-day median, 16 Nov 2026).
- Of the 3 benchmarks both models have, MiniMax M3.1 Flash leads 2 (BenchLeader, 16 Nov 2026).
- On Arena WebDev, MiniMax M3.1 Flash is rated 1584 and Qwen3.8 27B 1590 (LMArena, as published by 16 Nov 2026).
Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Qwen3.8 27B | 62 vs 55 category score; Arena WebDev 1590 vs 1584 | Medium |
| Agents & tool use | MiniMax M3.1 Flash | 55 vs 44 category score; Arena Agent -2.7 vs -4.8 | High |
| Reasoning | Qwen3.8 27B | 58 vs 50 category score | High |
| Images and other media | Either | 62 vs 61 category score | Either |
| Chat quality (human preference) | – | Only Qwen3.8 27B has an Arena rating (61 category score) | Not comparable |
| Lowest cost | MiniMax M3.1 Flash | $0.262 vs $1.06 per 1M blended; agentic coding session $0.011 vs $0.037 | – |
| Fastest first answer | Qwen3.8 27B | 0.7 s (Qwen3.8 27B) vs 43.3 s (MiniMax M3.1 Flash) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- MiniMax M3.1 Flash and Qwen3.8 27B are neck and neck on quality (54.3 vs 55.1, within the ±7.2 margin).
- MiniMax M3.1 Flash is stronger in agents & tools, multimodal.
- Qwen3.8 27B is stronger in coding, reasoning.
- MiniMax M3.1 Flash is 4.2× cheaper ($0.262 vs $1.09 per 1M blended).
- MiniMax M3.1 Flash streams 2.1× faster (151 vs 73 tokens per second).
At a glance
| Spec | MiniMax M3.1 Flash | Qwen3.8 27B |
|---|---|---|
| Developer | MiniMax | Alibaba |
| Released | 19 Oct 2026 | 14 Aug 2026 |
| Context window (tokens) | 1,000,000 | 262,144 |
| Compared at | MiniMax M3.1 Flash | Qwen3.8 27B |
| Open weights | Yes | Yes |
| Price in / out / cached input per 1M | $0.150 / $0.600 / $0.015 | $0.420 / $3.00 / $0.085 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | MiniMax M3.1 Flash | Qwen3.8 27B | Leader |
|---|---|---|---|
| Reasoning | 50.1 | 57.6 | Qwen3.8 27B |
| Coding | 55.4 | 61.6 | Qwen3.8 27B |
| Agents & tools | 55.2 | 44.2 | MiniMax M3.1 Flash |
| Maths | – | 54.6 | not comparable |
| Knowledge | 45.9 | – | not comparable |
| Human preference | – | 61.3 | not comparable |
| Multimodal | 61.5 | 60.7 | level |
| Long context | 64.5 | – | not comparable |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Maths, Knowledge, Instruction following, Long context, Human preference.
Where sources disagree
Across the 3 benchmarks both have, MiniMax M3.1 Flash leads 2, Qwen3.8 27B leads 1. Every overall index that rates both puts MiniMax M3.1 Flash ahead.
| Overall index | MiniMax M3.1 Flash | Qwen3.8 27B | Leader |
|---|---|---|---|
| BenchLeader Index | 54.3 | 55.1 | neck and neck |
| AA Intelligence Index v5.0.1 | 29.1 | 10.0† | MiniMax M3.1 Flash |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | MiniMax M3.1 Flash | Qwen3.8 27B |
|---|---|---|
| Blended price per 1M tokens | $0.262 | $1.06 |
| Input / output per 1M | $0.150 / $0.600 | $0.420 / $3.00 |
| Cached input per 1M | $0.015 | $0.085 |
| Price per task: chat reply | $0.0002 | $0.0011 |
| Price per task: summarize a 30-page report | $0.0022 | $0.0068 |
| Price per task: code edit | $0.0018 | $0.0070 |
| Price per task: agentic coding session | $0.011 | $0.037 |
| Price per task: structured extraction | $0.0004 | $0.0014 |
| Agentic coding session, input mostly cached | $0.0053 | $0.022 |
| Cost per benchmark task (AA Index) | $0.491 | – |
| Output tokens per task (AA Index) | 138k | – |
| Of which reasoning | 103k | – |
| Time per task (AA Index) | 15 min | – |
| Output speed | 151 tok/s | 73 tok/s |
| Time to first answer | 43.3 s | 0.7 s |
| Time to a complete answer, last measured 13 Nov 2026 | 141.5 s | – |
Each model is compared at its best-ranked reasoning-effort setting. MiniMax M3.1 Flash scores 54.3 and takes 43 s to its first answer token. Qwen3.8 27B scores 55.1 and takes 1 s to its first answer token; Qwen3.8 27B (xhigh) scores 52.2 in 109 s, Qwen3.8 27B (medium) scores 50.2 in 24 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
120 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | MiniMax M3.1 Flash | Qwen3.8 27B |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 54.3 | 55.1 |
| Agents & tools score | BenchLeader | 55.2 | 44.2 |
| Coding score | BenchLeader | 55.4 | 61.6 |
| Knowledge score | BenchLeader | 45.9 | – |
| Long context score | BenchLeader | 64.5 | – |
| Multimodal score | BenchLeader | 61.5 | 60.7 |
| Reasoning score | BenchLeader | 50.1 | 57.6 |
| Human preference score | BenchLeader | – | 61.3 |
| Maths score | BenchLeader | – | 54.6 |
| Blended price $/M | BenchLeader | $0.262 | $1.06 |
| Output speed | BenchLeader | 151 tok/s | 73 tok/s |
| Time to first answer | BenchLeader | 43.3 s | 0.7 s |
| Context window | BenchLeader | 1,000,000 tokens | 262,144 tokens |
| Reasoning | |||
| SimpleBench | SimpleBench | – | 60.2% |
| Arena Hard Prompts | Arena (LMArena) | – | 1459 |
| LiveBench 2.0 Reasoning | LiveBench | 82.9%† | 80.0%† |
| LiveBench 1.0 Reasoning | LiveBench | – | 80.0% |
| GPQA Diamond (AA) | Artificial Analysis | – | 90.5%† |
| Humanity's Last Exam (AA) | Artificial Analysis | 48.0% | 33.9%† |
| HLE-Diamond (AA) | Artificial Analysis | 28.1% | 15.3%† |
| GPQA Diamond (Vals) | Vals AI | – | 88.9%† |
| ARC-AGI-1 | ARC Prize | – | 87.5%† |
| ARC-AGI-2 | ARC Prize | – | 42.4%† |
| CritPt | Artificial Analysis | 22.3% | 5.4%† |
| CritPt v2 hard subset (AA) | Artificial Analysis | 32.5% | 4.0%† |
| Arena Expert | Arena (LMArena) | – | 1473 |
| LMCA | Epoch AI Benchmarking Hub | – | 41.4% |
| DTBench | Epoch AI Benchmarking Hub | – | 88.0% |
| Coding | |||
| SciCode | SciCode | – | 46.6%† |
| Arena Coding | Arena (LMArena) | – | 1496 |
| Arena WebDev | Arena (LMArena) | 1584 | 1590 |
| LiveBench 2.0 Coding | LiveBench | 68.8%† | 69.3%† |
| LiveBench 1.0 Coding | LiveBench | – | 75.7% |
| SciCode (AA) | Artificial Analysis | 53.4% | 46.6%† |
| LiveCodeBench | Vals AI | – | 84.0%† |
| IOI | Vals AI | – | 39.1%† |
| SWE-bench (Vals) | Vals AI | – | 86.0%† |
| Code Migration | Vals AI | – | 14.2%† |
Show all 120 benchmarks
| Metric | Publisher | MiniMax M3.1 Flash | Qwen3.8 27B |
|---|---|---|---|
| Coding | |||
| ProgramBench | Vals AI | – | 0.0%† |
| Vibe Code Bench v1.1 | Vals AI | 73.9%† | 64.8%† |
| Vibe Code Bench 1-100 v1.1 | Vals AI | 19.8%† | – |
| SWE-Bench Pro V2 | Scale AI SEAL | 96.6%† | – |
| SWE-Bench Pro V2 HARD | Scale AI SEAL | 84.3%† | – |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | – | 45.0%† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 25.8% | 5.6%† |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 79.8%† |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 30.3† | 43.8† |
| Design Arena Website | Design Arena | 1277 | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | 38.6% | 34.2%† |
| FrontierCode 2.0 (AA) | Artificial Analysis | 20.8% | 14.9%† |
| Agents & tools | |||
| APEX-Agents | Mercor | – | 47.5% |
| Arena Agent | Arena (LMArena) | -2.7 | -4.8 |
| LiveBench Long Agentic | LiveBench | 25.7%† | 15.9%† |
| LiveBench Browser | LiveBench | 80.6%† | 24.2%† |
| LiveBench Agentic Coding | LiveBench | – | 61.4% |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 58.4%† |
| HiL-Bench | Scale AI SEAL | 65.7% | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 47.9% | 46.0%† |
| GDPval-AA v2.1 | Artificial Analysis | 1465 | 1426† |
| τ³-Banking (AA) | Artificial Analysis | – | 48.0%† |
| CyberBench | Vals AI | – | 67.8%† |
| Finance Agent v2 | Vals AI | – | 48.5%† |
| Finance Agent v2.1 | Vals AI | 58.8%† | – |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | 7.5%† | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 11.3%† |
| Legal Research Bench | Vals AI | 34.1%† | 36.1%† |
| SkillsBench | Vals AI | – | 38.1%† |
| Tax Agent Bench | Vals AI | 32.1%† | 30.9%† |
| Terminal-Bench 4.0 (Vals) | Vals AI | 25.3%† | – |
| Terminal-Bench Science | Vals AI | – | 0.0%† |
| Arena Agent Task Completion | Arena (LMArena) | -0.7 | -2.7 |
| AutomationBench | Artificial Analysis | 72.5% | 48.2%† |
| GDP.pdf (AA) | Artificial Analysis | 20.8% | 16.4%† |
| EnterpriseOps-Gym | Artificial Analysis | – | 44.2%† |
| AA-Briefcase v1.1 | Artificial Analysis | 1474 | 1391† |
| GDPx | GDPx by SemiAnalysis | 4.0%† | – |
| Design Arena Agentic | Design Arena | 1252 | – |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 3.3% | 0.5%† |
| CWE-Bench (AA) | Artificial Analysis | – | 26.7%† |
| DeepSec-Bench (AA) | Artificial Analysis | – | 11.5%† |
| CyberGym end-to-end (AA) | Artificial Analysis | – | 0.0%† |
| MKA (AA) | Artificial Analysis | 33.5% | – |
| Maths | |||
| ProofBench | Vals AI | – | 16.0% |
| LiveBench Mathematics | LiveBench | – | 86.2% |
| Arena Maths | Arena (LMArena) | – | 1451 |
| Knowledge | |||
| LiveBench 2.0 Data Analysis | LiveBench | 71.4%† | 76.6%† |
| LiveBench Research | LiveBench | 76.7%† | 61.3%† |
| LiveBench 1.0 Data Analysis | LiveBench | – | 76.6% |
| AA-Omniscience v1.1 | Artificial Analysis | -6.4 | -6.4† |
| MMLU-Pro | Vals AI | – | 84.3%† |
| LegalBench | Vals AI | – | 82.4%† |
| TaxEval | Vals AI | – | 70.8%† |
| PRBench Finance | Scale AI SEAL | 45.6% | – |
| PRBench Legal | Scale AI SEAL | 45.1% | – |
| Excel Modeling Benchmark | Vals AI | – | 59.7%† |
| Excel Modeling Benchmark v1.1 | Vals AI | 67.8%† | – |
| MedCode | Vals AI | – | 28.7%† |
| MedScribe | Vals AI | – | 83.8%† |
| Access to Justice Benchmark | Vals AI | 73.3%† | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 45.1% | 17.3%† |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 6.3% | 71.4%† |
| Instruction following | |||
| LiveBench Language | LiveBench | – | 74.3% |
| LiveBench Instruction Following | LiveBench | – | 72.7% |
| Arena Instruction Following | Arena (LMArena) | – | 1419 |
| Human preference | |||
| Arena Text | Arena (LMArena) | – | 1434 |
| EQ-Bench Creative Writing v3 | EQ-Bench | – | 1671 |
| Arena Creative Writing | Arena (LMArena) | – | 1359 |
| Arena Multi-turn | Arena (LMArena) | – | 1436 |
| Arena Longer Queries | Arena (LMArena) | – | 1443 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | – | 1270 |
| LiveBench Design | LiveBench | 71.2%† | 72.4%† |
| MMMU-Pro | Artificial Analysis | 79.1% | 76.3%† |
| MMMU-Pro (Vals) | Vals AI | – | 83.9%† |
| MortgageTax | Vals AI | – | 64.9%† |
| SAGE | Vals AI | 85.3%† | – |
| GeoBench | GeoBench | – | 38.0% |
| VIALS-AA | Handshake AI | 29.8% | – |
| Code2Video-AA | Artificial Analysis | – | 949† |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 79.7% | 82.0%† |
| MLCR | Artificial Analysis | 16.7% | 22.8%† |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | – | 152.6 |
| LiveBench 2.0 | LiveBench | 63.7%† | 54.0%† |
| LiveBench 1.0 | LiveBench | – | 75.3% |
| AA Intelligence Index v5.0.1 | Artificial Analysis | 29.1 | 20.1† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 42.7 | 33.7† |
| AA Progress Index v1 | Artificial Analysis | 317.6 | 303.1† |
| Vals Index | Vals AI | 53.8† | – |
| Vals Finance Index | Vals AI | 55.9%† | – |
| Vals Router Index | Vals AI | 53.8% | – |
| Vals Legal Index | Vals AI | 67.1%† | – |
| AA Cyber Index | Artificial Analysis | – | 12.7† |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 52.1% | – |
| Metadata | |||
| AA Openness Index | Artificial Analysis | 27.8 | 38.9† |
Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- MiniMax M3.1 Flash: BenchLeader Index 53.7 → 54.3.
- Qwen3.8 27B: BenchLeader Index 53.2 → 55.1.
- Qwen3.8 27B: Qwen3.8 27B cached input price moved from $0.10/M to $0.085/M (15% down).
- Qwen3.8 27B: Qwen3.8 27B input price moved from $0.45/M to $0.42/M (7% down).
- MiniMax M3.1 Flash: BenchLeader Index 54.3 → 53.7.
- Qwen3.8 27B: BenchLeader Index 54.0 → 53.2.
- MiniMax M3.1 Flash: 1 score revision.
- Qwen3.8 27B: 2 new results (SimpleBench, GeoBench).
- MiniMax M3.1 Flash: 6 score revisions.
- Qwen3.8 27B: 5 score revisions.
Related comparisons
- MiniMax M3.1 Flash vs Claude Opus 5.5
- MiniMax M3.1 Flash vs GPT-6 Astra
- MiniMax M3.1 Flash vs Claude Fable 5.1
- MiniMax M3.1 Flash vs GPT-6.1 Sol
- Qwen3.8 27B vs Claude Opus 5
- MiniMax M3.1 Flash vs Qwen3.8 2.4T A95B
- MiniMax M3.1 Flash vs Gemini 4 Argon
- MiniMax M3.1 Flash vs GPT-5.6 Sol
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
MiniMax M3.1 Flash vs Qwen3.8 27B: questions
- Is MiniMax M3.1 Flash better than Qwen3.8 27B?
- MiniMax M3.1 Flash and Qwen3.8 27B are neck and neck on quality (54.3 vs 55.1, within the ±7.2 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
- MiniMax M3.1 Flash vs Qwen3.8 27B: which is cheaper, and will that change?
- MiniMax M3.1 Flash is cheaper today: $0.262 against $1.09 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is MiniMax M3.1 Flash or Qwen3.8 27B better for coding?
- Qwen3.8 27B scores higher in coding (62 vs 55 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1584 for MiniMax M3.1 Flash against 1590 for Qwen3.8 27B.
- Which is faster, MiniMax M3.1 Flash or Qwen3.8 27B?
- MiniMax M3.1 Flash streams ~2× faster (151 vs 73 output tokens per second, Artificial Analysis 7-day median).
- Is MiniMax M3.1 Flash or Qwen3.8 27B better for agents and tool use?
- MiniMax M3.1 Flash scores higher in agentic tasks (55 vs 44 on the category index, where 50 is average). On Arena Agent, the top shared agents & tools benchmark, it is -2.7 for MiniMax M3.1 Flash against -4.8 for Qwen3.8 27B.
More questions (4)
- Which is better for long documents?
- MiniMax M3.1 Flash accepts more context: 1,000,000 against 262,144 tokens for Qwen3.8 27B.
- Which has lower latency?
- Qwen3.8 27B answers first: 0.7 s against 43.3 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Do independent benchmarks agree on MiniMax M3.1 Flash vs Qwen3.8 27B?
- Across the 3 benchmarks both have, MiniMax M3.1 Flash leads 2, Qwen3.8 27B leads 1. Every overall index that rates both puts MiniMax M3.1 Flash ahead.
- When was MiniMax M3.1 Flash released?
- MiniMax M3.1 Flash was released on 19 Oct 2026.