ModelsCompareBest forBenchmarksStatusPricingAPI
Compare / head-to-head

Llama 4 MaverickvsPhi-4 Reasoning

Llama 4 Maverick leads 2 of 2 shared benchmarks. Llama 4 Maverick has the larger context window (1M tokens).

Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.

Shared benchmarks
2 – 0
Llama 4 Maverick leads
Cheaper per token
list price, input + output
Larger context
Llama 4 Maverick
1M tokens
Provider uptime (30d)
— · —
Meta · Microsoft
Llama 4 Maverick
Meta · released 2025-04-05
textvision
Context
1M
Max out
Input /1M
Output /1M
Cached /1M
Scores
2 · 2 core
Phi-4 Reasoning
Microsoft · released 2025-04-30
text
Context
33K
Max out
33K
Input /1M
Output /1M
Cached /1M
Scores
4 · 4 core
Quality

Benchmark matrix

BenchmarkLlama 4 MaverickPhi-4 ReasoningΔEdge
Reported by both · 2
GPQA Diamond69.8% 65.8% +4 ptLlama 4 Maverick
MMLU-Pro80.5% 74.3% +6.2 ptLlama 4 Maverick
Only Llama 4 Maverick reports · 0
Llama 4 Maverick reports nothing Phi-4 Reasoning does not.
Only Phi-4 Reasoning reports · 2
AIME 2025not reported62.9%
HumanEval+not reported92.9%
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Llama 4 Maverick minus Phi-4 Reasoning in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing

Side by side

SpecLlama 4 MaverickPhi-4 ReasoningEdge
Context window1M tokens33K tokensLlama 4 Maverick
Max output33K tokens
Input price / 1M
Output price / 1M
Cached input / 1M
Input + output / 1M
Lower is cheaper. List prices; batch, tool and regional fees excluded.
Modalitiestext · visiontextLlama 4 Maverick
Released2025-04-052025-04-30
Cited benchmark scores24
Reliability

Provider status

All providers
Built by Respan
Which one wins on your data?

Public benchmarks are a starting point. Run Llama 4 Maverick and Phi-4 Reasoning on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.