AI models

Model index

Every model we track: what it costs, how much context it takes, and where to watch it in action.

Prices and specs are read from the maker’s own pages, with the source and check date on every model page. The picks are editorial and labelled, and the numbers behind them are cited.

Last verified: October 24, 2026Model leaderboard How we rank

21 models

Sort by

Moonshot AI
Kimi K3.1 Pro
New
The 3T-class challenger that undercuts Opus
Coding · Reasoning · Vision · Agentic
Input
$2.00
Output
$10.00
Context
1M
Open weights (announced)Best for: Deep reasoning
Task cost: $0.20Open the model page →
October 23, 2026
MiniMax
MiniMax M3.1 Flash
New
1M context and open weights at a Chinese price
Agentic · Coding · Vision · Runs locally
Input
$0.15
Output
$0.60
Context
1M
Open weightsBest for: Self-hosting · High volume
Task cost: $0.014Open the model page →
October 19, 2026
DeepSeek
DeepSeek V4.1 Flash
NewBest budget workhorse
Open weights at a fraction of frontier prices
Agentic · Coding · Vision
Input
$0.30
Output
$1.20
Context
1M
Open weightsBest for: Self-hosting · High volume
Task cost: $0.027Open the model page →
October 17, 2026
NVIDIA
Nemotron 3.5 Super VL
New
NVIDIA’s open vision-language model with a 1M window
Vision · Agentic · Runs locally · Tool use
Input
$0.15
Output
$0.50
Context
1.05M
Open weightsBest for: Self-hosting · High volume
Task cost: $0.013Open the model page →
October 15, 2026
Google DeepMind
Gemini 4 Argon
New
Google’s flagship: 1M context and long decodes
Agentic · Coding · Reasoning · Vision
Input
$2.00
Output
$10.00
Context
1.05M
ProprietaryBest for: Long output
Task cost: $0.20Open the model page →
October 13, 2026
Anthropic
Claude Haiku 5.5
New
Cheap, fast, and good enough for the boring 80%
High volume · Vision · Tool use
Input
$0.10
Output
$0.50
Context
1M
ProprietaryBest for: High volume
Task cost: $0.010Open the model page →
October 7, 2026
Mistral AI
Mistral Large 4
New
Europe’s 1T challenger
Agentic · Coding · Reasoning · Vision
Input
$1.36
Output
$4.18
Context
1M
Open weights (announced)Best for: Value
Task cost: $0.11Open the model page →
October 6, 2026
OpenAI
GPT-6.1 Sol
NewBest value
Near-Astra quality at a fifth of the price
Coding · Agentic · Vision · Computer use
Input
$2.00
Output
$10.00
Context
1.05M
ProprietaryBest for: Value
Task cost: $0.20Open the model page →
September 29, 2026
Anthropic
Claude Sonnet 5.5
New
The speed/intelligence balance most teams should start with
Agentic · Coding · Vision · Tool use
Input
$2.00
Output
$10.00
Context
1M
ProprietaryBest for: Value
Task cost: $0.20Open the model page →
September 28, 2026
Anthropic
Claude Opus 5.5
Best overall
Anthropic's default for long-running agentic coding
Agentic · Coding · Reasoning · Tool use · Vision
Input
$4.00
Output
$20.00
Context
1M
ProprietaryBest for: General use
Task cost: $0.40Open the model page →
September 22, 2026
OpenAI
GPT-6 Luna
The cheapest model OpenAI ships
High volume · Vision · Tool use
Input
$0.10
Output
$0.50
Context
1.05M
ProprietaryBest for: High volume
Task cost: $0.010Open the model page →
September 22, 2026
xAI
Grok 4.7
xAI’s frontier model with a 500K context
Agentic · Coding · Reasoning · Vision
Input
$2.00
Output
$6.00
Context
500K
ProprietaryBest for: Value
Task cost: $0.16Open the model page →
September 21, 2026
OpenAI
GPT-6 Astra
OpenAI’s most capable model for demanding work
Agentic · Coding · Reasoning · Vision · Computer use
Input
$10.00
Output
$50.00
Context
1.05M
ProprietaryBest for: General use
Task cost: $1.00Open the model page →
September 3, 2026
Google DeepMind
Gemini 3.8 Flash
Google’s fast model for agents and long code
Agentic · Coding · Vision · Tool use
Input
$0.75
Output
$3.75
Context
1.05M
ProprietaryBest for: Long output
Task cost: $0.075Open the model page →
September 2, 2026
Meta
Muse Spark 1.3
Meta’s agentic coder with a 1M window
Agentic · Coding · Vision · Runs locally
Input
$1.25
Output
$4.25
Context
1.05M
Open weightsBest for: Value
Task cost: $0.11Open the model page →
September 2, 2026
Anthropic
Claude Fable 5.1
Premium reasoning when Opus is not enough
Reasoning · Agentic · Coding · Vision
Input
$10.00
Output
$50.00
Context
1M
ProprietaryBest for: Deep reasoning
Task cost: $1.00Open the model page →
September 1, 2026
Z.ai
GLM-5.3 Flash
Best open weights
The cheap multimodal GLM
Coding · Vision · Runs locally · High volume
Input
$0.15
Output
$0.50
Context
1M
Open weightsBest for: Self-hosting · High volume
Task cost: $0.013Open the model page →
August 26, 2026
Z.ai
GLM-5.3
The open-weight coder that keeps closing the gap
Coding · Reasoning · Runs locally
Input
$1.40
Output
$4.40
Context
1M
Open weightsBest for: Self-hosting
Task cost: $0.11Open the model page →
August 18, 2026
Alibaba (Qwen)
Qwen 3.8 27B
The local 27B that fits on a laptop
Runs locally · Vision · Coding · High volume
Input
$0.20
Output
$2.50
Context
1M
Open weightsBest for: Self-hosting · High volume
Task cost: $0.035Open the model page →
August 14, 2026
DeepSeek
DeepSeek V4 Pro
The Pro tier for hard reasoning, not everyday coding
Reasoning · Coding · Runs locally
Input
$0.66
Output
$1.98
Context
1M
Open weightsBest for: Self-hosting
Task cost: $0.053Open the model page →
August 13, 2026
Cursor
Composer 2.5
Cursor’s own coding model, built into the editor
Agentic · Coding
Price
Cursor plan
Context
200K
ProprietaryBest for: Value
Task cost: Cursor planOpen the model page →
May 18, 2026

Picks

How the picks work

Model leaderboard

Every pick on this page is an editorial opinion, not a benchmark result. The reason names a fact you can check (a price, a context window, a licence, or the vendor’s own positioning), and each model page shows the caveat that could change the decision.

Best overall: Claude Opus 5.5Best value: GPT-6.1 SolBest budget workhorse: DeepSeek V4.1 FlashBest open weights: GLM-5.3 Flash