Qwen3 Max vs Qwen3.5 397B-A17B vs Claude Haiku 4.5
Qwen3.5 397B-A17B comes out ahead, 65 to 58 and 50 on our weighted score, and it is the cheaper option too.
Alibaba (Qwen)Qwen3 Max
50/100- ECI142.4
- Price$1.20 / $6.00
- Context262K
- Our pickAlibaba (Qwen)
Qwen3.5 397B-A17B
65/100- ECI146.7
- Price$0.60 / $3.60
- Context262K
AnthropicClaude Haiku 4.5
58/100- ECI142.4
- Price$1.00 / $5.00
- Context200K
Qwen3.5 397B-A17B is our pick
Qwen3.5 397B-A17B is the better all-round choice, scoring 65/100 against Claude Haiku 4.5 (58) and Qwen3 Max (50). It leads on capability, price and inputs & features. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityQwen3.5 397B-A17BCapabilities Index (ECI): Qwen3.5 397B-A17B 146.7 · Claude Haiku 4.5 142.4 · Qwen3 Max 142.4
- Lowest priceQwen3.5 397B-A17BQwen3.5 397B-A17B $1.35 · Claude Haiku 4.5 $2.00 · Qwen3 Max $2.40 per 1M tokens (3:1 blend)
- Longest contextQwen3 Max and Qwen3.5 397B-A17BQwen3 Max 262,144 · Qwen3.5 397B-A17B 262,144 · Claude Haiku 4.5 200,000 tokens
- Widest inputsQwen3.5 397B-A17BQwen3 Max: Text · Qwen3.5 397B-A17B: Text, Images, Audio, Video · Claude Haiku 4.5: Text, Images, PDFs
- Self-hostingQwen3.5 397B-A17BPublishes downloadable weights
| Measure | Weight | Qwen3 Max | Qwen3.5 397B-A17B | Claude Haiku 4.5 |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 68 | 74 | 69 |
| Price | 25% | 32 | 44 | 36 |
| Inputs & features | 15% | 25 | 90 | 80 |
| Context window | 10% | 37 | 37 | 32 |
| Overall | 100% | 50/100 | 65/100 | 58/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Qwen3 MaxAlibaba (Qwen) | Qwen3.5 397B-A17BAlibaba (Qwen) | Claude Haiku 4.5Anthropic |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 142.4 | 146.7 (best) | 142.4 |
| ECI rank | #91 of 148 | #67 of 148 (best) | #90 of 148 |
| GPQA DiamondGraduate-level science questions | 72.6% | 86.4% (best) | 71.2% |
| FrontierMath Tiers 1–3Research-level mathematics | 19.0% | 31.2% (best) | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 73.3% | 88.9% (best) | 66.7% |
| SimpleQA VerifiedShort factual questions | 48.8% (best) | — | 13.2% |
| Price per million tokens | |||
| Input | $1.20 | $0.60 (best) | $1.00 |
| Output | $6.00 | $3.60 (best) | $5.00 |
| Cached input | — | — | $0.10 |
| Blended (3:1) | $2.40 | $1.35 (best) | $2.00 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Alibaba API | Official Alibaba API | Official Anthropic API |
| Limits | |||
| Context window | 262,144 tokens (best) | 262,144 tokens (best) | 200,000 tokens |
| Max output | 65,536 tokens (best) | 65,536 tokens (best) | 64,000 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | Yes |
| PDFs | No | No | Yes |
| Audio | No | Yes | No |
| Video | No | Yes | No |
| Reasoning | No | Yes | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | Yes |
| Availability | |||
| Weights | Proprietary | Open | Proprietary |
| API model ID | qwen3-max | qwen3.5-397b-a17b | claude-haiku-4-5 |
| API providers | 16 | 23 | 34 (best) |
| Released | Sep 23, 2025 | Feb 15, 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
- Qwen3.5 397B-A17B$13.20
- Claude Haiku 4.5$20.00
Which should you choose?
Which is better: Qwen3 Max, Qwen3.5 397B-A17B or Claude Haiku 4.5?
Qwen3.5 397B-A17B is the better all-round choice, scoring 65/100 against Claude Haiku 4.5 (58) and Qwen3 Max (50). It leads on capability, price and inputs & features. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, Qwen3 Max, Qwen3.5 397B-A17B or Claude Haiku 4.5?
Qwen3.5 397B-A17B is cheaper at $0.60 input / $3.60 output per million tokens (official Alibaba 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 $1.35 per million tokens for Qwen3.5 397B-A17B versus $2.00 for Claude Haiku 4.5 (1.5× as much) and $2.40 for Qwen3 Max (1.8× as much).
Which scores higher on benchmarks?
Qwen3.5 397B-A17B scores higher on the Capabilities Index (ECI): Qwen3.5 397B-A17B 146.7 (#67 of 148), Claude Haiku 4.5 142.4 (#90 of 148) and Qwen3 Max 142.4 (#91 of 148). Their confidence ranges do not overlap (144.8–148.2 vs 139.5–144.3), so the gap is a real one. On individual benchmarks: GPQA Diamond — Qwen3.5 397B-A17B 86.4%, Qwen3 Max 72.6%, Claude Haiku 4.5 71.2%; OTIS Mock AIME 2024–2025 — Qwen3.5 397B-A17B 88.9%, Qwen3 Max 73.3%, Claude Haiku 4.5 66.7%.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3 Max, Qwen3.5 397B-A17B and Claude Haiku 4.5 yet, so there is no like-for-like coding score. On overall capability, Qwen3.5 397B-A17B 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 and Qwen3.5 397B-A17B have the largest context windows (262,144 and 262,144 tokens), against 200,000 for Claude Haiku 4.5. Maximum output per response: Qwen3 Max up to 65,536, Qwen3.5 397B-A17B up to 65,536, Claude Haiku 4.5 up to 64,000 tokens.
Which can read images, PDFs, audio or video?
Qwen3 Max accepts text; Qwen3.5 397B-A17B accepts text, images, audio and video; Claude Haiku 4.5 accepts text, images and PDFs. Qwen3.5 397B-A17B handles the widest range of inputs.
Are any of these open source?
Qwen3.5 397B-A17B publishes its weights and can be self-hosted; Qwen3 Max and Claude Haiku 4.5 is proprietary.
Which is newer?
Qwen3.5 397B-A17B is the newest, released Feb 15, 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.
