Best AI Model by Use Case
Skip the benchmarks: here's the best model for what you actually want to do.
Computed live from our benchmark database (Arena Elo, SWE-bench, GPQA, hallucination rates, measured speed, image and video arena Elo). Updated Oct 31, 2026.
</>Coding & Software Engineering
Coding overall
Claude Fable 5.181.6 AA Coding Index: highest on Artificial Analysis's own coding runs
Runner-up: GPT-6 Astra
Autonomous coding agent
Kimi K393.4% SWE-bench Verified: resolves the most real GitHub issues
Runner-up: Ember-1
Competitive programming
OrcaCyber Nano 1.092.2% LiveCodeBench: strongest on fresh contest problems
Runner-up: Gemini 3.1 Pro
Fast iterative coding
Kimi K2 Thinking85.3% LiveCodeBench at 120 tok/s: quality at interactive speed
Runner-up: LFM2.5-2.6B
AaWriting & Content
Long-form writing
Gemini 4 Carbon1528 Arena Elo: most preferred by human voters in blind tests
Runner-up: Gemini 4 Argon
Copywriting & marketing
Muse Wonder1507 Arena Elo at 143 tok/s: top-tier prose, fast enough to iterate
Runner-up: Gemini 3.8 Flash
Translation & multilingual
Gemini 3.1 Pro91.0% MMLU-Pro: broadest cross-language knowledge
Runner-up: Sakana Namazu
Faithful summarization
Gemini 4 Argon15.1% hallucination rate: least likely to invent facts
Runner-up: MiniMax-M3
?Research & Analysis
Scientific research
GPT-6 Astra96.0% GPQA Diamond: best PhD-level science reasoning
Runner-up: Gemini 3.1 Pro
Math & quantitative analysis
Inkling97.1% AIME 2026: top competition-math accuracy
Runner-up: AliceAI-Foundation-80B-A3B-Base
Grounded answers with citations
Gemini 4 Argon15.1% hallucination rate with 1520 Arena Elo: accurate and well-liked
Runner-up: MiniMax-M3
Document & chart analysis
Qwen3.6 35B A3B81.7% MMMU: strongest visual reasoning over documents and figures
Runner-up: Seed-2.0-Mini
[]Image & Video Generation
Text to image
GPT Image 2.5 Sunburst (max)1222 Elo in Artificial Analysis's text-to-image arena (blind human votes, as of Oct 31, 2026)
Runner-up: GPT Image 2.5 Flare (max)
Image editing
GPT Image 2.5 Sunburst (max)1199 Elo in Artificial Analysis's image editing arena (blind human votes, as of Oct 31, 2026)
Runner-up: MAI-Image-2.7-Preview
Text to video
HeyGen Video 11143 Elo in Artificial Analysis's text-to-video arena (blind human votes, as of Oct 31, 2026)
Runner-up: MiniMax H3 Max
Image to video
Utopai X1155 Elo in Artificial Analysis's image-to-video arena (blind human votes, as of Oct 31, 2026)
Runner-up: Dreamina Seedance 2.5
$Speed & Budget
Fastest inference
Nemotron 3.5 Super VL474 tok/s measured throughput (Artificial Analysis)
Runner-up: Trinity Large Thinking
Quick Q&A / chatbot
Muse Spark 1.21492 Arena Elo at 264 tok/s: quality answers without the wait
Runner-up: GPT-5.4 mini
How we pick these recommendations
Every pick on this page is computed from our benchmark database each time you load it, the same data behind the Full Benchmark Leaderboard, where each score links to its source: Artificial Analysis, Vals AI, LMArena, Vectara, llm-stats, or the developer's own announcement. Each task ranks models by the metric that matters for it:
- Coding: SWE-bench Verified (real GitHub issues) and LiveCodeBench (fresh contest problems)
- Writing: LMArena Elo (blind human preference) and measured speed
- Research: GPQA Diamond, AIME, and hallucination rate
- Trustworthiness: Artificial Analysis AA-Omniscience hallucination rate (lower is better)
- Speed & budget: measured tokens/sec and per-token pricing
- Image & video generation: Elo from Artificial Analysis's image and video arenas, where people vote between two outputs without knowing which model made them
Also see: Best Coding LLM, Best Reasoning LLM, Best Cheap LLM, Full Benchmark Leaderboard.