Agentic
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 61.6
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
MiniMax M2.5 vs Qwen3.6 Plus
Updated August 18, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
Qwen3.6 Plus has the higher public score estimate, 64.79 versus 58.68, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
1 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Long documents
Prompts that approach the documented context limit
Qwen3.6 Plus
Qwen3.6 Plus has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. MiniMax M2.5 does not fit this workload in one request. MiniMax M2.5 has no published cached-input rate, so cached tokens use its listed input rate. Qwen3.6 Plus has no comparable published API token rate.
Confidence: rate-fallback
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
1 Shared
43 Qwen3.6 Plus only
- Shared results
- 1
- MiniMax M2.5 only
- 0
- Qwen3.6 Plus only
- 43
- Like-for-like categories
- 0 / 8
Category results, on a stated basis
Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.
Coding
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 70.3
- Weighted basis
- 0 vs 2 rows
- Reading
- Not comparable
Reasoning
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 62.0
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
Knowledge
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 57.1
- Weighted basis
- 0 vs 4 rows
- Reading
- Not comparable
Math
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 60.5
- Weighted basis
- 0 vs 4 rows
- Reading
- Not comparable
Multilingual
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 84.7
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
Multimodal
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 79.8
- Weighted basis
- 0 vs 2 rows
- Reading
- Not comparable
Instruction following
Not comparable- MiniMax M2.5
- Not measured
- Qwen3.6 Plus
- 82.3
- Weighted basis
- 0 vs 2 rows
- Reading
- Not comparable
| Category | MiniMax M2.5 | Qwen3.6 Plus | Weighted basis | Reading |
|---|---|---|---|---|
| Agentic | Not measured | 61.6 | Not comparable0 vs 1 rows | Not comparable |
| Coding | Not measured | 70.3 | Not comparable0 vs 2 rows | Not comparable |
| Reasoning | Not measured | 62.0 | Not comparable0 vs 1 rows | Not comparable |
| Knowledge | Not measured | 57.1 | Not comparable0 vs 4 rows | Not comparable |
| Math | Not measured | 60.5 | Not comparable0 vs 4 rows | Not comparable |
| Multilingual | Not measured | 84.7 | Not comparable0 vs 1 rows | Not comparable |
| Multimodal | Not measured | 79.8 | Not comparable0 vs 2 rows | Not comparable |
| Instruction following | Not measured | 82.3 | Not comparable0 vs 2 rows | Not comparable |
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
A shared-evidence shape is not available.
BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- MiniMax M2.5
- $0.0009
- Fits in one request
- Qwen3.6 Plus
- API rate not published
- Fits in one request
Qwen3.6 Plus has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- MiniMax M2.5
- $0.0186
- Fits in one request
- Qwen3.6 Plus
- API rate not published
- Fits in one request
Qwen3.6 Plus has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- MiniMax M2.5
- $0.078
- Does not fit in one request
- Cached input priced at the published list-input rate
- Qwen3.6 Plus
- API rate not published
- Fits in one request
- Cached-input rate unavailable
MiniMax M2.5 does not fit this workload in one request. MiniMax M2.5 has no published cached-input rate, so cached tokens use its listed input rate. Qwen3.6 Plus has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationMiniMax M2.5Qwen3.6 Plus
Context window
Maximum documented context; output-token limits may be lower.
MiniMax M2.5
128K
Qwen3.6 Plus
1M
API model ID
MiniMax M2.5
Not sourced
Qwen3.6 Plus
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
MiniMax M2.5
Not published
Qwen3.6 Plus
No comparable hosted API rate
Documented inputs
MiniMax M2.5
Not sourced
Qwen3.6 Plus
Not sourced
Documented outputs
MiniMax M2.5
Not sourced
Qwen3.6 Plus
Not sourced
Provider availability
MiniMax M2.5
Not sourced
Qwen3.6 Plus
Not sourced
Reasoning profile
MiniMax M2.5
Non-Reasoning
Qwen3.6 Plus
Reasoning
Weight access
MiniMax M2.5
Proprietary
Qwen3.6 Plus
Proprietary
License
MiniMax M2.5
Proprietary
Qwen3.6 Plus
Proprietary
Release date
MiniMax M2.5
2025-10-01
Qwen3.6 Plus
2026-04-02
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- Qwen3.6 Plus has the higher public score estimate, 64.79 versus 58.68, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Qwen3.6 Plus has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence44 rows
Agentic
Terminal-Bench 2.0
MiniMax M2.5—Qwen3.6 Plus61.6%SourceNot directly comparable
Claw-Eval
MiniMax M2.5—Qwen3.6 Plus58.8%SourceNot directly comparable
QwenClawBench
MiniMax M2.5—Qwen3.6 Plus57.2%SourceNot directly comparable
τ³-bench results
MiniMax M2.5—Qwen3.6 Plus70.7%SourceNot directly comparable
VITA-Bench
MiniMax M2.5—Qwen3.6 Plus44.3%SourceNot directly comparable
DeepPlanning
MiniMax M2.5—Qwen3.6 Plus41.5%SourceNot directly comparable
Toolathlon
MiniMax M2.5—Qwen3.6 Plus39.8%SourceNot directly comparable
MCP Atlas
MiniMax M2.5—Qwen3.6 Plus48.2%SourceNot directly comparable
MCP-Tasks
MiniMax M2.5—Qwen3.6 Plus74.1%SourceNot directly comparable
WideResearch
MiniMax M2.5—Qwen3.6 Plus74.3%SourceNot directly comparable
Gert Labs
MiniMax M2.5—Qwen3.6 Plus50.60%SourceNot directly comparable
ResearchClawBench
MiniMax M2.5—Qwen3.6 Plus18.0%SourceNot directly comparable
Coding
Vibe Code Bench
Shared sourceMiniMax M2.514.85%Qwen3.6 Plus25.56%Qwen3.6 Plus leads this result
SWE-bench Verified
MiniMax M2.5—Qwen3.6 Plus78.8%SourceNot directly comparable
SWE-bench Pro
MiniMax M2.5—Qwen3.6 Plus56.6%SourceNot directly comparable
SWE Multilingual
MiniMax M2.5—Qwen3.6 Plus73.8%SourceNot directly comparable
LiveCodeBench v6
MiniMax M2.5—Qwen3.6 Plus87.1%SourceNot directly comparable
EEBench
MiniMax M2.5—Qwen3.6 Plus12.7%SourceNot directly comparable
Reasoning
Knowledge
GPQA
MiniMax M2.5—Qwen3.6 Plus90.4%SourceNot directly comparable
SuperGPQA
MiniMax M2.5—Qwen3.6 Plus71.6%SourceNot directly comparable
MMLU-Pro
MiniMax M2.5—Qwen3.6 Plus88.5%SourceNot directly comparable
MMLU-Redux
MiniMax M2.5—Qwen3.6 Plus94.5%SourceNot directly comparable
C-Eval
MiniMax M2.5—Qwen3.6 Plus93.3%SourceNot directly comparable
HLE
MiniMax M2.5—Qwen3.6 Plus28.8%SourceNot directly comparable
Math
AIME26
MiniMax M2.5—Qwen3.6 Plus95.3%SourceNot directly comparable
HMMT Feb 2025
MiniMax M2.5—Qwen3.6 Plus96.7%SourceNot directly comparable
HMMT Nov 2025
MiniMax M2.5—Qwen3.6 Plus94.6%SourceNot directly comparable
HMMT Feb 2026
MiniMax M2.5—Qwen3.6 Plus87.8%SourceNot directly comparable
MMAnswerBench
MiniMax M2.5—Qwen3.6 Plus83.8%SourceNot directly comparable
FrontierMath v2 (Tiers 1-3)
MiniMax M2.5—Qwen3.6 Plus26.207%SourceNot directly comparable
FrontierMath v2 (Tier 4)
MiniMax M2.5—Qwen3.6 Plus8.333%SourceNot directly comparable
Multilingual
Multimodal
MMMU
MiniMax M2.5—Qwen3.6 Plus86.0%SourceNot directly comparable
MMMU-Pro
MiniMax M2.5—Qwen3.6 Plus78.8%SourceNot directly comparable
MathVision
MiniMax M2.5—Qwen3.6 Plus88.0%SourceNot directly comparable
VideoMMMU
MiniMax M2.5—Qwen3.6 Plus84.0%SourceNot directly comparable
ScreenSpot Pro
MiniMax M2.5—Qwen3.6 Plus68.2%SourceNot directly comparable
CharXiv
MiniMax M2.5—Qwen3.6 Plus81.5%SourceNot directly comparable
V*
MiniMax M2.5—Qwen3.6 Plus96.9%SourceNot directly comparable
Instruction following
Frequently asked questions
Which is better, MiniMax M2.5 or Qwen3.6 Plus?
Qwen3.6 Plus has the higher public score estimate, 64.79 versus 58.68, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, MiniMax M2.5 or Qwen3.6 Plus?
The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.
Which is better for agentic tasks, MiniMax M2.5 or Qwen3.6 Plus?
The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.
Which costs less, MiniMax M2.5 or Qwen3.6 Plus?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, MiniMax M2.5 or Qwen3.6 Plus?
Qwen3.6 Plus has the larger documented context window: 1M, compared with 128K.
Related comparisons
Last updated August 18, 2026
Watch MiniMax M2.5 vs Qwen3.6 Plus
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.