LLMs 101

Monthly model tracker

How AI models rank when
real humans compare them

Benchmark scores don't tell the full story. This tracker translates human preference data and real-world usage into plain-English tiers — updated every month.

Updated 22nd July 2026
🏆
What is human preference ranking?
Real users are shown two anonymous AI responses to the same prompt and pick the one they prefer. Thousands of these battles produce an Elo-style ranking — the same system used to rank chess players.
💰
What does "cost vibe" mean?
Instead of listing raw dollar-per-token figures (which change constantly), we use plain-English cost tiers: Free, Low, Standard, and Premium — reflecting how expensive it is to run each model at real production scale.
📊
Why not show benchmark scores?
Benchmarks like MMLU or HumanEval measure narrow academic tasks. A score of "86.2%" means nothing in practice. Human preference rankings and real use-case guidance are far more useful for most readers.
🔄
How often is this updated?
Monthly. The AI landscape moves fast — a model released last month can jump several tiers. Check back each month, or read the Trends section for major shifts as they happen.
Show
How to read this table: Rankings reflect our monthly editorial review of current AI benchmarks and independent testing, combined with qualitative assessment of real-world professional use. Tier 1 = consistently preferred across that review. Tiers are relative — all listed models are genuinely excellent by any historical standard.
1
OpenAI
GPT-5.6 Sol
Flagship frontier intelligence model
🏆 Tier 1 - Top
Best for
The hardest reasoning, coding, and agentic work. Runs behind ChatGPT Pro mode and powers long autonomous tasks split across subagents.
Premium Closed API
2
Anthropic
Claude Opus 4.8
Flagship intelligence model
🥈 Tier 1 - Top
Best for
The most demanding coding, judgment, and agentic tasks. Leads on hard coding and is tuned for honesty and catching its own mistakes.
Premium Closed API
3
Google DeepMind
Gemini 3.1 Pro
Flagship reasoning and multimodal model
🥉 Tier 1 - Top
Best for
Complex projects needing deep reasoning plus strong multimodal understanding. Google's top generally-available Pro model while 3.5 Pro remains delayed.
Premium Closed API
4
xAI
Grok 4.5
Opus-class coding and agentic model
Tier 1 - Top (challenger)
Best for
Software engineering and agentic workflows, trained alongside Cursor on real developer sessions. Fast and token-efficient with live access to X data.
Standard Closed API
5
Anthropic
Claude Sonnet 5
Best-value all-rounder
Tier 2 - Best value Claude
Best for
Everyday coding, writing, and agentic work at near-Opus quality for roughly 40% less. The default model on Claude Free and Pro plans.
Standard Closed API
6
Google DeepMind
Gemini 3.6 Flash
Fast, efficient workhorse model
Tier 2 - Best value Gemini
Best for
High-volume coding, knowledge work, and multimodal tasks. Delivers quality close to Gemini Pro using up to 17% fewer tokens than 3.5 Flash.
Low Closed API
7
Moonshot AI
Kimi K3
Largest open-weight frontier model
Tier 3 - Open-weight frontier
Best for
Long-horizon coding, knowledge work, and reasoning at open-weight cost. At 2.8T parameters it benchmarks near the very top of widely available models.
Low Open weights
8
Z.ai
GLM-5.2
Top open-weight coding model
Tier 3 - Open-weight value
Best for
Long-horizon agentic coding on a 1M-token context at a fraction of closed-model cost. MIT-licensed and downloadable, it lands near Opus 4.8 on key agent benchmarks.
Low Open weights
9
DeepSeek
DeepSeek V4-Pro
Cost-efficient open reasoning model
Tier 3 - Open-weight value
Best for
Reasoning, math, and agentic coding on a 1M-token context. MIT-licensed open weights that rival top closed models while leading most other open releases.
Ultra-low Open weights
10
OpenAI
GPT-5.6 Luna
Budget high-volume model
Tier 4 - Speed and cost leader
Best for
Cost-sensitive, high-volume workloads on OpenAI's current frontier generation. The cheapest sibling of the GPT-5.6 family, built for scale.
Low Closed API
11
Anthropic
Claude Haiku 4.5
Fast, low-cost small model
Tier 4 - Speed and cost leader
Best for
Real-time chat, customer support, and high-volume subagent tasks. Delivers near-frontier coding at low cost and high speed.
Low Closed API
12
Mistral
Mistral Medium 3.5
European balanced enterprise model
Tier 2 - Emerging challenger
Best for
Balanced enterprise workloads from Europe's leading lab, with strong performance and simplified deployment. A credible sovereign alternative to US-based models.
Standard Closed API

* Llama models are free to download and run but require your own hardware or cloud compute. Self-hosting costs (electricity, GPU rental) vary. · Rankings reflect monthly editorial review of current AI benchmarks and independent testing, combined with editorial assessment — see our Methodology & Glossary page for full detail. · Cost tiers are illustrative and based on that same monthly review — actual pricing changes frequently; check each provider's current pricing page for exact figures. · This tracker focuses on general-purpose chat and reasoning models. Specialised models (image generation, audio, video) are not included.