Across 10 comparable benchmarks, DeepSeek-V3-0324 has better reported results on 8 and worse on 2.
Strong evidence10 of 10 shared results countedHow we compare
Price basis: Listed rates vs listed rates. The price section below shows the applicable periods.
| Fact | DeepSeek-V3-0324 | Llama 4 Maverick |
|---|---|---|
| Context window | 128K tokens | 1M tokens |
| Input / 1M | $1.14 | — |
| Output / 1M | $4.56 | — |
| Blended (4:1) | $1.82 | — |
| Status | Available | Available |
| Released | March 24, 2025 | April 5, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 4 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
DeepSeek-V3-0324 ahead on 2 of 2 counted
Split 1–1 of 2 counted
DeepSeek-V3-0324 ahead on 1 of 1 counted
No fixed-scale average — see the rows in Benchmark results
DeepSeek-V3-0324 ahead on 1 of 1 counted
DeepSeek-V3-0324 ahead on 1 of 1 counted
No fixed-scale average — see the rows in Benchmark results
No shared comparable results for Mathematics, Computer & tool use, Research & search, Long context, Cybersecurity.
Published scores, with matching modes where available. Gaps show DeepSeek-V3-0324 relative to Llama 4 Maverick. Reading the results
| Benchmark | DeepSeek-V3-0324 | Llama 4 Maverick | Gap |
|---|---|---|---|
| Aider-PolyglotAgentic Development · Accuracy | 55.1Normal · no tools | 15.6Normal · no tools | +39.5 pts#19 vs #42 of 47 |
| Terminal Bench HardAgentic Development · Accuracy | 15.2Normal · tools | 6.8Normal · tools | +8.4 pts#107 vs #123 of 157 |
| ARC-AGI-1Abstract Generalization · Score (%) | 9.0Normal · no tools | 4.4≠Thinking · no tools | +4.6 pts#82 vs #85 of 87 |
| SimpleBenchCommonsense · Score (AVG@5) | 27.2Normal · no tools | 27.7Normal · no tools | 0.98×#90 vs #88 of 106 |
| SciCodeScientific Computing · Score | 39.0Normal · no tools | 31.7Normal · no tools | 1.23×#77 vs #91 of 98 |
| GDP.pdfDocuments & Charts · Task pass rate (%) | 1.6Normal · no tools | 0.8Normal · no tools | +0.8 pts#86 vs #93 of 96 |
| τ³-BankingService Workflows · Score | 4.7Normal · tools | 3.7Normal · tools | 1.27×#113 vs #116 of 120 |
| HLEKnowledge Exams · Accuracy | 5.2Normal · no tools | 5.7≠Reported best | −0.5 pts#143 vs #139 of 151 |
| MMLUKnowledge Exams · Accuracy | 86.5Normal · no tools | 85.5Normal · no tools | +1.0 pts#25 vs #34 of 114 |
| MMLU-ProKnowledge Exams · Accuracy | 81.2Normal · no tools | 62.9Normal · no tools | +18.3 pts#54 vs #108 of 153 |
Agentic Development · Accuracy
Agentic Development · Accuracy
Abstract Generalization · Score (%)
Commonsense · Score (AVG@5)
Scientific Computing · Score
Documents & Charts · Task pass rate (%)
Service Workflows · Score
A dash means no result is available.
| Benchmark | DeepSeek-V3-0324 | Llama 4 Maverick |
|---|---|---|
| AIME 2024 | 59.4 | — |
| AIME2025 | 47.7 | — |
| FrontierMath | — | 0.7 |
| GPQA Diamond | 68.4 | — |
| GSM8K | 96.3 | — |
| IMO-ProofBench | 4.3 | — |
| MATH | — | 61.2 |
| MATH-500 | 94.0 | — |
| IOI 2024 (Vals v1) | 1.7 | — |
| IOI 2025 (Vals v1) | 1.7 | — |
| LiveCodeBench | 49.2 | — |
| MBPP | — | 77.6 |
| SWE-bench Verified | 38.8 | — |
| Terminal-Bench 1.0 | 13.3 | — |
| Pinch Bench | — | 46.1 |
| ECI | — | 132 |
| Creative Writing | 1472 | — |
| DROP | 89.7 | — |
| MMMU | — | 73.4 |
| MMMU-Pro | — | 62.1 |
| τ²-Bench | 38.8 | — |
| ARC-AGI-2 | — | 0.0 |
| GAIA | 29.4 | — |
| SimpleQA | 27.2 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
DeepSeek-V3-0324: Listed text rates; pricing tier not confirmed.
Llama 4 Maverick: Listed text rates; pricing tier not confirmed.
| Price / 1M tokens | DeepSeek-V3-0324 | Llama 4 Maverick | Ratio |
|---|---|---|---|
| Input | $1.14 | — | — |
| Output | $4.56 | — | — |
| Cached input | — | — | — |
| Cache write | — | — | — |
| Blended | $1.82 | — | — |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Lower prices are highlighted. The cache hit rate applies to input tokens only; models without a published cached-input rate use their normal input price for those tokens.Effective dates: DeepSeek-V3-0324 2026-11-14; Llama 4 Maverick —.Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
| Attribute | DeepSeek-V3-0324 | Llama 4 Maverick |
|---|---|---|
| Availability | Available | Available |
| Provider | DeepSeek-AI | Facebook AI Research Lab |
| Model type | Chat model | Multimodal model |
| First release | March 24, 2025 | April 5, 2025 |
| Context window | 128K tokens | 1M tokens |
| Max output | — | 4K tokens |
| Total parameters | 671B | 400B |
| Active parameters | 37B | — |
| Architecture | Mixture of Experts (MoE) | Dense |
| Documented capabilities | Multilingual | Multilingual |
| Input modalities | Text | Text, Image, Audio, Video |
| Output modalities | Text | Text |
| Weights | Open weights | Open weights |
| Licence | MIT License | Llama4 License |
| Commercial use | Commercial use allowed | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used DeepSeek-V3-0324 or Llama 4 Maverick? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.