Get started

Llama 4 Maverick

Timed at 39.1 tok/s, scored 16.3 on the Coding Index, on 6 Design Arena boards, and with results on 1 openly licensed board.

retrieved sha256 · 0bb53a77713f

Models/Meta Llama/Llama 4 MaverickUSD per 1M tokens · as OpenRouter routes it
meta-llama/llama-4-maverickMeta Llama · base listing · Llama4 tokenizer
OpenRouter list price · the routed rate, as the catalog carried itretrieved 17 Nov 16:29 UTC from openrouter.ai ↗
Input$0.35per 1M tokens
Output$1.00per 1M tokens
Cached input$0.17per 1M tokens
Context524k524,288 tokens
What Vitna read · 2 providers, each from its own pricing page, kept as dated recordsin / out per 1M tokens
AWSopen weights$0.24 / $0.97confirmed 22 Nov 2026
Parasailopen weights$0.35 / $1.00confirmed 22 Nov 2026

The same weights served by 2 providers, each at its own list price. OpenRouter's rate is its routed rate across whichever of them it uses, and it need not match any one of them.

What the list says · the fields OpenRouter publishes for this id10 fields
Listed
Canonical idmeta-llama/llama-4-maverick-17b-128e-instruct
Context524,288 tokens
Max output32,768 tokens
Modalitiestext, image → text
TokenizerLlama4
Moderationnot moderated by OpenRouter
Knowledge cutoff2024-08-31
What the weights say · read from the model's own repository on Hugging Faceretrieved from huggingface.co ↗
Licencellama4, tagged other
Parameters402B · 401,583,781,376 declared
Model classLlama4ForConditionalGeneration
Downloadgated, each request approved by the maker
Published

Measured speed

10,565 requests · 2 endpoints · 30-minute window
Throughput39.1 tok/sMedian output rate, weighted across the 2 endpoints serving it by the requests each answered.the speed board
First token751msMedian wait before the first token, over the same window and weighted the same way.how Vitna reads a figure
Endpoint spread39 to 111.5The slowest and the fastest route serving this model, in tokens per second. Speed is a property of the route, not of the weights.the board, fastest first

OpenRouter's own timing of its routed traffic, weighted by requests, over one 30-minute window read . Every route is on the endpoint ledger below.

Every endpoint

1 priced · 2 timed · cheapest first
  1. Parasailparasail/fp8 · listed$0.35$1.00$0.17fp833k39 tok/s24751ms10,555 req100%
  2. Googleno priceno priceno priceno offerno offer112 tok/s70.3729ms10 reqno offer

Each track runs from a route's p90, the rate nine in ten of its requests met or beat, up to its median. Weighted by requests, the figure above reports where the traffic went rather than what one route does: 1 of the 2 timed routes are slower than it, and the middle route runs at 75.3 tok/s. The widest median-to-p90 gap is Parasail's, 1.63x.

Uptime is OpenRouter's own figure for each seller over the trailing day, read once a day, and the column prints the latest reading, taken . Every one of the 2 sellers with a reading was at 99% or better on each of the 23 days held, 23 Oct 2026 to 21 Nov 2026. Source: OpenRouter's public endpoints API, one reading a day for each seller: 1,211 sellers carried one on the latest of the 23 days held.

Source: OpenRouter's public endpoints API. 1257 sellers across 359 models when the mirror ran, 186 of them served by more than one. List prices, not what anybody paid; a discount is the seller's own. Read . Speed is OpenRouter's own timing over one 30-minute window, read .

What each seller accepts

16 parameters · 1 seller
  1. frequency_penaltyevery seller
  2. logit_biasevery seller
  3. logprobsevery seller
  4. max_tokensevery seller
  5. presence_penaltyevery seller
  6. repetition_penaltyevery seller
  7. response_formatevery seller
  8. seedevery seller
  9. stopevery seller
  10. structured_outputsevery seller
  11. temperatureevery seller
  12. tool_choiceevery seller
  13. toolsevery seller
  14. top_kevery seller
  15. top_logprobsevery seller
  16. top_pevery seller

The one seller that publishes a list accepts these 16. Each list is the seller's own declaration in OpenRouter's endpoints API, read , and nothing here tested one.

Price position

342 priced models

Blended $0.513 per million tokens, 3 in to 1 out: cheaper than 211 of the 342 priced models in the catalog. Each bar is one stretch of the catalog on a logarithmic price scale; the mark is this listing.

Measured quality

145 models with an index · 132 on a design arena board

Each mark is one board this model stands on, placed at the share of that board it scores above, and led to its number in the list under the rule. The columns and the ledger below carry the figures.

Intelligence Indexnot scoredArtificial Analysis has not scored this model on its Intelligence Index.the intelligence leaderboard
Coding Index16.3Position 101 of 118, above 14% of the models scored.the coding leaderboard
Agentic Index0.6Position 90 of 94, above 3% of the models scored.the agentic leaderboard
  1. 3D scenes · Models881 Elowin rate 40.2% · source rank 132the board, 113 ranked
  2. Code · Models868 Elowin rate 35.8% · source rank 143the board, 120 ranked
  3. Data visualisation · Models882 Elowin rate 38.4% · source rank 133the board, 110 ranked
  4. Game development · Models827 Elowin rate 33.7% · source rank 145the board, 120 ranked
  5. UI components · Models888 Elowin rate 40.8% · source rank 138the board, 116 ranked
  6. Websites · Models862 Elowin rate 34.4% · source rank 147the board, 126 ranked
Epoch Capabilities Index · a fit by a third party over every benchmark this model has results onthe whole index, with intervals
Index132.2129.8 to 134.3
Position154 of 258above 40% of the models placed
Results behind it13benchmark results
As Epoch names itLlama 4 Maverickreleased 6 Apr 2025

Position 154 of the 258 models Epoch has placed, which is not the population of either board above, on a scale anchored at Claude 3.5 Sonnet 130 and GPT-5 150. Epoch states that absolute values are meaningless by themselves and that comparisons between models are what the index is for, that values are not linearly related to benchmark accuracy, and that a highly specialised model may score low while being capable in its own domain.

Benchmark results

1 of 6 boards · 1 result
Aider polyglot · submitted, Apache-2.01 of 1 task · 69 runs scored on the task · effort unstated
TaskScore
polyglot pass rate15.60

Percent of 225 Exercism exercises solved with a second attempt allowed, higher is better. The id carries the edit format, because the publisher says a model alone does not identify a run. Aider polyglot names this run Llama 4 Maverick (whole), with no effort setting written on it. Read from raw.githubusercontent.com ↗, which dates its own data . Aider polyglot benchmark, Apache-2.0, by the Aider project. Results are submitted by contributors rather than produced by the publisher.

1 of the 6 openly licensed boards Vitna mirrors carries a result for Llama 4 Maverick, each printed as its publisher published it and none added to another. Epoch AI Benchmarking Hub, MTEB, LiveBench, HELM and Vectara hallucination leaderboard have not published one under a name that matches its listing, and names are matched to the catalog exactly and never guessed. A task that is a link opens its column on /capability, with every run scored on it. Every board, with its licence and what its publisher measures, is on /benchmarks.

Sources

Source: OpenRouter (openrouter.ai/models), as of 17 Nov 16:29 UTC. sha256 0bb53a77713f over 853,270 bytes.
Catalog: OpenRouter's public model list.
Intelligence, coding and agentic indexes: Artificial Analysis, as published by OpenRouter.
Elo, win rate and rank: Design Arena, as published by OpenRouter.
Source: each model's own Hugging Face repository (huggingface.co), as of 20 Nov 2026. 162 repositories read, 4 did not answer.
Licence, parameter count, model class and gating: each model's own Hugging Face repository, the one OpenRouter names for the listing.
A licence is the tag the uploader put on the repository, read from nothing else and verified by nobody.
A parameter count is the total the repository's safetensors files declare; for a mixture of experts that is every expert, not the share one token runs through.
Source: the Epoch AI Capabilities Index (epoch.ai/eci), as of 20 Nov 2026. 258 models placed over 2917 benchmark results.
Epoch AI Benchmarking Hub data, by Epoch AI, under CC BY 4.0.
Vitna ran none of these evaluations and adjusted none of these figures.