Hermes 4 405B vs Qwen3 VL 235B A22B Thinking vs Relace Search
Qwen3 VL 235B A22B Thinking comes out ahead, 44 to 35 and 35 on our weighted score, and it is the cheaper option too.
Nous ResearchHermes 4 405B
35/100- ECI—
- Price$1.00 / $3.00
- Context131K
- Our pick
Alibaba (Qwen)Qwen3 VL 235B A22B Thinking
44/100- ECI—
- Price$0.50 / $3.95
- Context131K
Relace
Relace Search
35/100- ECI—
- Price$1.00 / $3.00
- Context256K
Qwen3 VL 235B A22B Thinking is our pick
Qwen3 VL 235B A22B Thinking is the better all-round choice, scoring 44/100 against Hermes 4 405B (35) and Relace Search (35). It leads on price and inputs & features. Relace Search wins on context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceQwen3 VL 235B A22B ThinkingQwen3 VL 235B A22B Thinking $1.36 · Hermes 4 405B $1.50 · Relace Search $1.50 per 1M tokens (3:1 blend)
- Longest contextRelace SearchRelace Search 256,000 · Hermes 4 405B 131,072 · Qwen3 VL 235B A22B Thinking 131,072 tokens
- Widest inputsQwen3 VL 235B A22B ThinkingHermes 4 405B: Text · Qwen3 VL 235B A22B Thinking: Text, Images · Relace Search: Text
- Self-hostingHermes 4 405B and Qwen3 VL 235B A22B ThinkingPublishes downloadable weights
| Measure | Weight | Hermes 4 405B | Qwen3 VL 235B A22B Thinking | Relace Search |
|---|---|---|---|---|
| Price | 50% | 40 | 42 | 40 |
| Inputs & features | 30% | 35 | 60 | 25 |
| Context window | 20% | 25 | 25 | 36 |
| Overall | 100% | 35/100 | 44/100 | 35/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Hermes 4 405BNous Research | Qwen3 VL 235B A22B ThinkingAlibaba (Qwen) | Relace SearchRelace |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $1.00 | $0.50 (best) | $1.00 |
| Output | $3.00 (best) | $3.95 | $3.00 (best) |
| Cached input | — | — | — |
| Blended (3:1) | $1.50 | $1.36 (best) | $1.50 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 5 providers | Median of 5 providers | Median of 3 providers |
| Limits | |||
| Context window | 131,072 tokens | 131,072 tokens | 256,000 tokens (best) |
| Max output | 117,964 tokens | 32,768 tokens | 128,000 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | Yes | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | Open | Proprietary |
| API model ID | — | — | — |
| API providers | 5 (best) | 5 (best) | 3 |
| Released | Aug 26, 2025 | Sep 23, 2025 | Dec 8, 2025 |
| Knowledge cutoff | Aug 31, 2024 | Mar 31, 2025 | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Hermes 4 405B$16.00
- Qwen3 VL 235B A22B Thinking$12.90
- Relace Search$16.00
Which should you choose?
Which is better: Hermes 4 405B, Qwen3 VL 235B A22B Thinking or Relace Search?
Qwen3 VL 235B A22B Thinking is the better all-round choice, scoring 44/100 against Hermes 4 405B (35) and Relace Search (35). It leads on price and inputs & features. Relace Search wins on context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Hermes 4 405B, Qwen3 VL 235B A22B Thinking or Relace Search?
Qwen3 VL 235B A22B Thinking is cheaper at $0.50 input / $3.95 output per million tokens (median across 5 API providers). Hermes 4 405B costs $1.00 input / $3.00 output per million tokens (median across 5 API providers); Relace Search costs $1.00 input / $3.00 output per million tokens (median across 3 API providers). At a typical mix of three input tokens to one output token, that is $1.36 per million tokens for Qwen3 VL 235B A22B Thinking versus $1.50 for Hermes 4 405B (1.1× as much) and $1.50 for Relace Search (1.1× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Hermes 4 405B has not been scored yet, Qwen3 VL 235B A22B Thinking has not been scored yet and Relace Search has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Hermes 4 405B, Qwen3 VL 235B A22B Thinking and Relace Search yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Relace Search has the largest context window at 256,000 tokens, against 131,072 for Hermes 4 405B and 131,072 for Qwen3 VL 235B A22B Thinking. Maximum output per response: Hermes 4 405B up to 117,964, Qwen3 VL 235B A22B Thinking up to 32,768, Relace Search up to 128,000 tokens.
Which can read images, PDFs, audio or video?
Hermes 4 405B accepts text; Qwen3 VL 235B A22B Thinking accepts text and images; Relace Search accepts text. Qwen3 VL 235B A22B Thinking handles the widest range of inputs.
Are any of these open source?
Hermes 4 405B and Qwen3 VL 235B A22B Thinking publishes its weights and can be self-hosted; Relace Search is proprietary.
Which is newer?
Relace Search is the newest, released Dec 8, 2025. Qwen3 VL 235B A22B Thinking came out Sep 23, 2025; Hermes 4 405B came out Aug 26, 2025. Knowledge cutoff: Hermes 4 405B Aug 31, 2024, Qwen3 VL 235B A22B Thinking Mar 31, 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 Nov 2, 2026.
