Hermes 4 - Llama-3.1 405B
Good value at this level: no ranked model within 2 Index points costs three quarters of the price or less.
No cheaper ranked model scores higher: it is on the price frontier.
As of 16 Nov 2026, Hermes 4 - Llama-3.1 405B by Nous Research is not yet ranked on the BenchLeader Index, which needs at least five independent results across two categories. API pricing: $1.00 per 1M input tokens and $3.00 per 1M output tokens ($1.50 blended at 3:1), list price on 16 Nov 2026. It streams about 34 tokens/s (Artificial Analysis). Released 27 Aug 2025.
Key facts (as of 16 Nov 2026)
- Hermes 4 - Llama-3.1 405B API pricing: $1.00 per 1M input tokens and $3.00 per 1M output tokens ($1.50 blended at 3:1) (list price, 16 Nov 2026).
- Hermes 4 - Llama-3.1 405B streams about 34 tokens/s (Artificial Analysis, 16 Nov 2026).
- Hermes 4 - Llama-3.1 405B was released on 27 Aug 2025 (Nous Research).
In detail. Hermes 4 - Llama-3.1 405B is a Nous Research open-weights model, released 27 Aug 2025. It is not yet ranked: the index needs at least five independent results across two categories. At $1.50 per million tokens blended it is pricier than most ranked models. Output speed of 34 tokens per second puts it in the slowest quarter, with a first answer in 5.1 s. It has been measured at 2 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 27 Aug 2025.stale: no new result in six months
- Blended price
- $1.50/M
- $1.00 in · $3.00 out
- via Artificial Analysis
- Output speed
- 34 tok/s
- measured by Artificial Analysis
- First answer
- 5.07 s
- first token 5.07 s
- Context
- 128k
- via Artificial Analysis
- Full answer
- 36 s
- last measured 13 Nov 2026
- Cost per task
- –
- Tokens per task
- –
- Released
- 27 Aug 2025
- Agents & tools42
- Instruction following38
- Long context34
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoningbest | – | – | 34 tok/s | 5.07 s | $0.0013 | Agents & tools 42 · Instruction following 38 · Long context 34 |
| thinking | – | – | 38 tok/s | 22 s | $0.0013 | Agents & tools 44 · Instruction following 36 · Long context 35 |
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Metadata
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| AA Openness Indexnot in index | 47.2#92 | 47.2#92 | Artificial Analysis |
Reasoning
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| GPQA Diamond (AA)not in index | 53.6%#417 | 72.7%#280 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.2%#545 | 10.9%#327 | Artificial Analysis |
| CritPtnot in index | 0.0%#337 | 0.3%#291 | Artificial Analysis |
Coding
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| LiveCodeBench (AA)not in index | 54.6%#124 | 68.6%#76 | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| Terminal-Bench Hard | 9.8%#233 | 11.4%#220 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 26.6%#285 | 22.2%#319 | Artificial Analysis |
Maths
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| AIME 2025 (AA)not in index | 15.3%#207 | 69.7%#93 | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| IFBench | 34.8%#318 | 32.7%#343 | Artificial Analysis |
Long context
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| AA-LCR v1.1 | 22.0%#448 | 22.3%#446 | Artificial Analysis |
Composite
| Benchmark | no reasoning | thinking | Source |
|---|---|---|---|
| AA Intelligence Index v4.3.2not in index | 7.4#487 | 7.5#482 | Artificial Analysis |
| AA Progress Index v1not in index | 181.2#477 | 206.8#395 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0013 | – | 13.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.014 | – | 22.7 s |
| Code edit | 6,000 / 1,500 | $0.011 | – | 49.2 s |
| Agentic coding session | 60,000 / 4,000 | $0.072 | – | 2.0 min |
| Structured extraction | 2,000 / 200 | $0.0026 | – | 11.0 s |
More from Nous Research
See also
Questions people ask
- Why does Hermes 4 - Llama-3.1 405B have no BenchLeader Index score?
- The index needs at least five independent results across at least two categories, and as of 16 Nov 2026 Hermes 4 - Llama-3.1 405B does not have them yet. Launch-post scores are never used, so it is ranked once independent publishers have measured it.
- How much does Hermes 4 - Llama-3.1 405B cost per 1M tokens?
- As of 16 Nov 2026, Hermes 4 - Llama-3.1 405B costs $1.00 per 1M input tokens and $3.00 per 1M output tokens, $1.50 blended (three input tokens per output token).
- When was Hermes 4 - Llama-3.1 405B released?
- Nous Research released Hermes 4 - Llama-3.1 405B on 27 Aug 2025. BenchLeader has results for it at 2 reasoning-effort settings.
Data as of 16 Nov 2026. Compare these configurations.
Cite as: BenchLeader, “Hermes 4 - Llama-3.1 405B: benchmarks, pricing, speed and rank”, data as of 16 Nov 2026, https://www.benchleader.com/models/hermes-4-llama-3-1-405b.