Qwen3-Next 80B-A3B (Thinking) vs Kimi K2 Instruct 0905 vs Hermes 4 405B
No independent capability benchmark covers every model here yet. The score uses price, inputs & features and context window.
Price in rating: Included · Exclude price
Alibaba (Qwen)
Qwen3-Next 80B-A3B (Thinking)
37/100- ECI—
- Price$0.50 / $6.00
- Context262K
Moonshot AI
Kimi K2 Instruct 0905
39/100- ECI—
- Price$0.60 / $2.50
- Context262K
Nous Research
Hermes 4 405B
37/100- ECI—
- Price$1.00 / $3.00
- Context131K
Capability evidence is incomplete
There is not enough shared capability evidence to name an overall winner. The scores (Kimi K2 Instruct 0905 39/100, Hermes 4 405B 37/100, Qwen3-Next 80B-A3B (Thinking) 37/100) compare the available specifications and price. 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 priceKimi K2 Instruct 0905Kimi K2 Instruct 0905 $1.07 · Hermes 4 405B $1.50 · Qwen3-Next 80B-A3B (Thinking) $1.88 per 1M tokens (3:1 blend)
- Longest contextQwen3-Next 80B-A3B (Thinking) and Kimi K2 Instruct 0905Qwen3-Next 80B-A3B (Thinking) 262,144 · Kimi K2 Instruct 0905 262,144 · Hermes 4 405B 131,072 tokens
- Widest inputsSame inputsQwen3-Next 80B-A3B (Thinking): Text · Kimi K2 Instruct 0905: Text · Hermes 4 405B: Text
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Qwen3-Next 80B-A3B (Thinking) | Kimi K2 Instruct 0905 | Hermes 4 405B |
|---|---|---|---|---|
| Price | 50% | 35 | 46 | 40 |
| Inputs & features | 30% | 35 | 25 | 35 |
| Context window | 20% | 43 | 43 | 32 |
| Specs & price only | 100% | 37/100 | 39/100 | 37/100 |
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 | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.50 (best) | $0.60 | $1.00 |
| Output | $6.00 | $2.50 (best) | $3.00 |
| Cached input | — | — | — |
| Blended (3:1) | $1.88 | $1.07 (best) | $1.50 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Alibaba API | Median of 13 providers | Median of 4 providers |
| Limits | |||
| Context window | 262,144 tokens (best) | 262,144 tokens (best) | 131,072 tokens |
| Max output | 32,768 tokens | 98,304 tokens | 117,964 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | No | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | OpenModified MIT | Open |
| API model ID | qwen3-next-80b-a3b-thinking | — | — |
| API providers | 6 | 14 (best) | 4 |
| Released | Sep 2025 | Sep 5, 2025 | Aug 26, 2025 |
| Knowledge cutoff | Apr 2025 | — | Aug 31, 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
Qwen3-Next 80B-A3B (Thinking)$17.00
Kimi K2 Instruct 0905$11.00
Hermes 4 405B$16.00
Which should you choose?
Which is better: Qwen3-Next 80B-A3B (Thinking), Kimi K2 Instruct 0905 or Hermes 4 405B?
There is not enough shared capability evidence to name an overall winner. The scores (Kimi K2 Instruct 0905 39/100, Hermes 4 405B 37/100, Qwen3-Next 80B-A3B (Thinking) 37/100) compare the available specifications and price. 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, Qwen3-Next 80B-A3B (Thinking), Kimi K2 Instruct 0905 or Hermes 4 405B?
Kimi K2 Instruct 0905 is cheaper at $0.60 input / $2.50 output per million tokens (median across 13 API providers). Hermes 4 405B costs $1.00 input / $3.00 output per million tokens (median across 4 API providers); Qwen3-Next 80B-A3B (Thinking) costs $0.50 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 $1.07 per million tokens for Kimi K2 Instruct 0905 versus $1.50 for Hermes 4 405B (1.4× as much) and $1.88 for Qwen3-Next 80B-A3B (Thinking) (1.7× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Qwen3-Next 80B-A3B (Thinking) has not been scored yet, Kimi K2 Instruct 0905 has not been scored yet and Hermes 4 405B has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3-Next 80B-A3B (Thinking), Kimi K2 Instruct 0905 and Hermes 4 405B 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?
Qwen3-Next 80B-A3B (Thinking) and Kimi K2 Instruct 0905 have the largest context windows (262,144 and 262,144 tokens), against 131,072 for Hermes 4 405B. Maximum output per response: Qwen3-Next 80B-A3B (Thinking) up to 32,768, Kimi K2 Instruct 0905 up to 98,304, Hermes 4 405B up to 117,964 tokens.
Which can read images, PDFs, audio or video?
Qwen3-Next 80B-A3B (Thinking) accepts text; Kimi K2 Instruct 0905 accepts text; Hermes 4 405B accepts text. They handle the same number of input types.
Are any of these open source?
Yes, all three publish their weights (Modified MIT), so you can self-host them.
Which is newer?
Kimi K2 Instruct 0905 is the newest, released Sep 5, 2025. Qwen3-Next 80B-A3B (Thinking) came out Sep 2025; Hermes 4 405B came out Aug 26, 2025. Knowledge cutoff: Qwen3-Next 80B-A3B (Thinking) Apr 2025, Hermes 4 405B Aug 31, 2024.
How do you decide the winner?
This rating uses price (50%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (30%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (20%, 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 18, 2026.