Lowest typical cost
Gemini 3.1 Flash-Lite
$0.56 / 1M total
Model comparisons
Compare skills, cost, and fit.
Compare coding, reasoning, math, instruction-following, speed, and context. Then check price and value before choosing what to test.
Show company
Capability signals
Compare skills and behavior
Compare by
Building, editing, and debugging software
RankModelRelative signalScore
1
Claude Opus 5
Anthropic
97
2
GPT-5.6 Sol
OpenAI
96
3
Claude Opus 4.8
Anthropic
94
4
Claude Sonnet 5
Anthropic
94
5
Grok Build 0.1
xAI
92
6
GPT-5.6 Terra
OpenAI
91
7
Claude Sonnet 4.6
Anthropic
91
8Google
Gemini 2.5 Pro
91
9
Grok 4.5
xAI
89
10
Grok 4.3
xAI
86
11
GPT-5.6 Luna
OpenAI
84
12
Claude Haiku 4.5
Anthropic
84
13Google
Gemini 3.6 Flash
84
14Google
Gemini 3.1 Flash-Lite
78
These are directional AI Stack composite signals on a normalized 0–100 scale, not official provider benchmark scores. Use them to narrow the field, then test finalists on your own work.
Price and capacity
Compare cost, value, and working memory
Rank by
ModelGood forInput / 1MOutput / 1MTypical costOutput / $1Memory
1Google
Gemini 3.1 Flash-Lite
Low cost · High volume
$0.25$1.50$0.56667K1M2
Grok Build 0.1
xAI
Coding · Implementation
$1.00$2.00$1.25500K256K3
Grok 4.3
xAI
Value · Large context
$1.25$2.50$1.56400K1M4
Claude Haiku 4.5
Anthropic
Speed · High volume
$1.00$5.00$2.00200K200K5
GPT-5.6 Luna
OpenAI
High volume · Fast iteration
$1.00$6.00$2.25167K1.05M6Google
Gemini 3.6 Flash
Speed · Multimodal
$1.50$7.50$3.00133K1M7
Grok 4.5
xAI
Reasoning · Current information
$2.00$6.00$3.00167K500K8Google
Gemini 2.5 Pro
Complex reasoning · Coding
$1.25$10.00$3.44100K1.05M9
Claude Sonnet 5
Anthropic
Coding · Data analysis
$2.00$10.00$4.00100K1M10
GPT-5.6 Terra
OpenAI
Balanced value · Coding
$2.50$15.00$5.6367K1.05M11
Claude Sonnet 4.6
Anthropic
Debugging · Implementation
$3.00$15.00$6.0067K1M12
Claude Opus 4.8
Anthropic
Long tasks · Coding
$5.00$25.00$10.0040K1M13
Claude Opus 5
Anthropic
Agentic coding · Long tasks
$5.00$25.00$10.0040K1M14
GPT-5.6 Sol
OpenAI
Complex coding · Agents
$5.00$30.00$11.2533K1.05M15
Claude Fable 5
Anthropic
Deep reasoning · Long-horizon agents
$10.00$50.00$20.0020K1MInput is what you send; output is what the model writes back. Typical cost estimates one million total tokens as 75% input and 25% output. API prices can vary with caching, batching, long prompts, tools, and service tiers.
Most output per dollar
Gemini 3.1 Flash-Lite
667K tokens / $1
Largest working memory
GPT-5.6 Sol
1.05M