models.fyiPreview data

Llama 4 Maverick

MetaOpen weights

Meta's open non-reasoning model: no effort dial at all. Cheap, fast, fine for everyday text — a baseline for what 'no thinking' buys you.

Index @ Standard

52.4

Context

1M

$/M in · out

$0.27 · $0.85

Median speed

160 tok/s

Released

Apr 5, 2025

02·Which effort for which task

The sweet spot, per task

Cheapest effort tier that keeps ≥96% of this model's peak quality on the benchmarks behind each task type. Below the floor number, quality is not okay — it just fails quietly.

Everyday writing & Q&A

Standard

Chat, drafting, summaries, general questions.

keeps 100.0% of peakpeak tier needed

Code assist

Standard

Autocomplete, single functions, small fixes.

keeps 100.0% of peakpeak tier needed

Coding agents

Standard

Real repos, multi-file changes, end-to-end tickets.

keeps 100.0% of peakpeak tier needed

Math & quantitative

Standard

Derivations, statistics, financial models.

keeps 100.0% of peakpeak tier needed

Hard research problems

Standard

Novel, frontier, genuinely difficult questions.

keeps 100.0% of peakpeak tier needed

Agents & tool use

Standard

Long workflows with tools, policies, and state.

keeps 100.0% of peakpeak tier needed

03·The effort ladder

Quality and cost, tier by tier

Intelligence Index by effort

Cost per typical task by effort

04·Benchmark detail

Scores at Standard

at Standardat Standard (peak)

Benchmark	Standard	Peak	What it predicts
MMLU-Pro	80.0	80.0	Everyday Q&A · Drafting emails & docs · Summarizing articles
GPQA Diamond	70.0	70.0	Technical analysis · Root-cause debugging · Scientific writing
Humanity's Last Exam	9.0	9.0	Novel research questions · Frontier problem solving
AIME 2025	40.0	40.0	Financial modeling · Statistics · Algorithm design
SWE-bench Verified	44.0	44.0	Coding agents · Bug fixing in real repos · Multi-file refactors
LiveCodeBench	48.0	48.0	Writing functions from scratch · Algorithms · Code autocomplete
τ²-bench	52.0	52.0	Tool-using agents · Customer-facing automation · Multi-step workflows
IFEval	82.0	82.0	Structured output (JSON) · Templated generation · Batch pipelines

05·Host variants

Same weights, different model

Hosts serve Llama 4 Maverick at different quantizations and stacks. Index below is measured per host at Standard effort — the headline number only applies to the reference serving.

Host	Quant	Index	Δ vs reference	$/M in · out	tok/s	TTFT	Uptime	Gateway id
Together AIreference	fp8	52.4	—	$0.27 · $0.85	160	0.30s	99.2%	llama-4-maverick
Groq	int8	50.8	-1.6	$0.20 · $0.60	550	0.20s	98.9%	llama-4-maverick@groq
DeepInfra	fp8	51.9	-0.5	$0.17 · $0.60	140	0.45s	98.5%	llama-4-maverick@deepinfra

06·Drift

Is it still the same model this week?

Together AIGroqDeepInfra

Weekly Intelligence Index · dashed rules mark serving-change events

No serving-change events detected for Llama 4 Maverick in the tracked window.

07·Call it

One key, this model, your effort

The gateway speaks the OpenAI Chat Completions dialect. Pick a host with llama-4-maverick@host syntax, set reasoning_effort, and the sweet spot above becomes one line of config. Full docs →

chat with Llama 4 Maverick

curl https://models.fyi/api/v1/chat/completions \
  -H "Authorization: Bearer $MFYI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-4-maverick",
    "reasoning_effort": "standard",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
# local dev: replace https://models.fyi with http://localhost:3000