Qwen3 Max vs MiniMax-M2.5 vs Claude Haiku 4.5
Too close to call on our weighted score (MiniMax-M2.5 61, Claude Haiku 4.5 58, Qwen3 Max 50). The right pick depends on what you value most.
Alibaba (Qwen)Qwen3 Max
50/100- ECI142.4
- Price$1.20 / $6.00
- Context262K
MiniMaxMiniMax-M2.5
61/100- ECI146.7
- Price$0.30 / $1.20
- Context205K
AnthropicClaude Haiku 4.5
58/100- ECI142.4
- Price$1.00 / $5.00
- Context200K
Too close to call
It is close. Our weighted score puts them within 3 points (MiniMax-M2.5 61/100, Claude Haiku 4.5 58/100, Qwen3 Max 50/100), so choose by what matters most for your work: MiniMax-M2.5 for raw capability and Qwen3 Max for long inputs. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityMiniMax-M2.5Capabilities Index (ECI): MiniMax-M2.5 146.7 · Claude Haiku 4.5 142.4 · Qwen3 Max 142.4
- Lowest priceMiniMax-M2.5MiniMax-M2.5 $0.525 · Claude Haiku 4.5 $2.00 · Qwen3 Max $2.40 per 1M tokens (3:1 blend)
- Longest contextQwen3 MaxQwen3 Max 262,144 · MiniMax-M2.5 204,800 · Claude Haiku 4.5 200,000 tokens
- Widest inputsClaude Haiku 4.5Qwen3 Max: Text · MiniMax-M2.5: Text · Claude Haiku 4.5: Text, Images, PDFs
- Self-hostingMiniMax-M2.5Publishes downloadable weights
| Measure | Weight | Qwen3 Max | MiniMax-M2.5 | Claude Haiku 4.5 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 68 | 74 | 69 |
| Price | 25% | 32 | 63 | 36 |
| Inputs & features | 15% | 25 | 35 | 80 |
| Context window | 10% | 37 | 32 | 32 |
| Overall | 100% | 50/100 | 61/100 | 58/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Qwen3 MaxAlibaba (Qwen) | MiniMax-M2.5MiniMax | Claude Haiku 4.5Anthropic |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 142.4 | 146.7 (best) | 142.4 |
| ECI rank | #91 of 148 | #66 of 148 (best) | #90 of 148 |
| GPQA DiamondGraduate-level science questions | 72.6% (best) | — | 71.2% |
| FrontierMath Tiers 1–3Research-level mathematics | 19.0% | — | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 73.3% (best) | — | 66.7% |
| SimpleQA VerifiedShort factual questions | 48.8% (best) | — | 13.2% |
| Price per million tokens | |||
| Input | $1.20 | $0.30 (best) | $1.00 |
| Output | $6.00 | $1.20 (best) | $5.00 |
| Cached input | — | $0.03 (best) | $0.10 |
| Blended (3:1) | $2.40 | $0.525 (best) | $2.00 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Alibaba API | Official MiniMax (minimax.io) API | Official Anthropic API |
| Limits | |||
| Context window | 262,144 tokens (best) | 204,800 tokens | 200,000 tokens |
| Max output | 65,536 tokens | 131,072 tokens (best) | 64,000 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | Yes |
| PDFs | No | No | Yes |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | Yes | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | Yes |
| Availability | |||
| Weights | Proprietary | Open | Proprietary |
| API model ID | qwen3-max | MiniMax-M2.5 | claude-haiku-4-5 |
| API providers | 16 | 21 | 34 (best) |
| Released | Sep 23, 2025 | Feb 12, 2026 | Oct 15, 2025 |
| Knowledge cutoff | Apr 2025 | — | Feb 28, 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Qwen3 Max$24.00
- MiniMax-M2.5$5.40
- Claude Haiku 4.5$20.00
Which should you choose?
Which is better: Qwen3 Max, MiniMax-M2.5 or Claude Haiku 4.5?
It is close. Our weighted score puts them within 3 points (MiniMax-M2.5 61/100, Claude Haiku 4.5 58/100, Qwen3 Max 50/100), so choose by what matters most for your work: MiniMax-M2.5 for raw capability and Qwen3 Max for long inputs. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Qwen3 Max, MiniMax-M2.5 or Claude Haiku 4.5?
MiniMax-M2.5 is cheaper at $0.30 input / $1.20 output per million tokens (official MiniMax (minimax.io) API price). Claude Haiku 4.5 costs $1.00 input / $5.00 output per million tokens (official Anthropic API price); Qwen3 Max costs $1.20 input / $6.00 output per million tokens (official Alibaba API price). At a typical mix of three input tokens to one output token, that is $0.525 per million tokens for MiniMax-M2.5 versus $2.00 for Claude Haiku 4.5 (3.8× as much) and $2.40 for Qwen3 Max (4.6× as much).
Which scores higher on benchmarks?
MiniMax-M2.5 scores higher on the Capabilities Index (ECI): MiniMax-M2.5 146.7 (#66 of 148), Claude Haiku 4.5 142.4 (#90 of 148) and Qwen3 Max 142.4 (#91 of 148). The confidence ranges of the top two overlap (142.3–147.9 vs 139.5–144.3), so treat the gap as small.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3 Max, MiniMax-M2.5 and Claude Haiku 4.5 yet, so there is no like-for-like coding score. On overall capability, MiniMax-M2.5 leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.
Which has the bigger context window?
Qwen3 Max has the largest context window at 262,144 tokens, against 204,800 for MiniMax-M2.5 and 200,000 for Claude Haiku 4.5. Maximum output per response: Qwen3 Max up to 65,536, MiniMax-M2.5 up to 131,072, Claude Haiku 4.5 up to 64,000 tokens.
Which can read images, PDFs, audio or video?
Qwen3 Max accepts text; MiniMax-M2.5 accepts text; Claude Haiku 4.5 accepts text, images and PDFs. Claude Haiku 4.5 handles the widest range of inputs.
Are any of these open source?
MiniMax-M2.5 publishes its weights and can be self-hosted; Qwen3 Max and Claude Haiku 4.5 is proprietary.
Which is newer?
MiniMax-M2.5 is the newest, released Feb 12, 2026. Claude Haiku 4.5 came out Oct 15, 2025; Qwen3 Max came out Sep 23, 2025. Knowledge cutoff: Qwen3 Max Apr 2025, Claude Haiku 4.5 Feb 28, 2025.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
