Tool
Which GenAI model fits which work? Price against position, by kind of work. Prices come from each provider’s own list; the position is our assessment. Both are accounted for below.
General purpose Knowledge work Software Customer interaction Documents and data Autonomous workflows Media and content
Strong and inexpensive GPT-6 Astra, $20, position 95 of 100. GPT-5.5 Pro, $68, position 92 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 76 of 100. GPT-5.4 Mini, $1.7, position 65 of 100. GPT-5.6 Luna, $0.45, position 60 of 100. GPT-5 Nano, $0.14, position 43 of 100. Claude Fable 5.1, $20, position 94 of 100. Claude Opus 5, $10, position 90 of 100. Claude Sonnet 5, $4, position 81 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 83 of 100. Gemini 3.8 Flash, $1.5, position 71 of 100. Gemini 3.5 Flash, $3.4, position 63 of 100. Gemini 3.5 Flash-Lite, $0.85, position 54 of 100. Gemini 2.5 Pro, $3.4, position 74 of 100. Gemini 2.5 Flash-Lite, $0.17, position 47 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-6 Astra GPT-5.6 Sol GPT-5.3 Codex GPT-5.4 Mini GPT-5.6 Luna GPT-5 Nano Claude Opus 5 Claude Sonnet 5 Gemini 3.8 Flash Gemini 3.5 Flash-Lite Gemini 2.5 Flash-Lite
Strong and inexpensive GPT-6 Astra, $20, position 99 of 100. GPT-5.5 Pro, $68, position 95 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 76 of 100. GPT-5.4 Mini, $1.7, position 65 of 100. GPT-5.6 Luna, $0.45, position 60 of 100. GPT-5 Nano, $0.14, position 34 of 100. Claude Fable 5.1, $20, position 96 of 100. Claude Opus 5, $10, position 90 of 100. Claude Sonnet 5, $4, position 81 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 83 of 100. Gemini 3.8 Flash, $1.5, position 71 of 100. Gemini 3.5 Flash, $3.4, position 63 of 100. Gemini 3.5 Flash-Lite, $0.85, position 48 of 100. Gemini 2.5 Pro, $3.4, position 74 of 100. Gemini 2.5 Flash-Lite, $0.17, position 39 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-6 Astra GPT-5.5 Pro GPT-5.6 Sol GPT-5.3 Codex GPT-5.4 Mini GPT-5.6 Luna GPT-5 Nano Claude Opus 5 Claude Sonnet 5 Gemini 3.8 Flash Gemini 3.5 Flash-Lite Gemini 2.5 Flash-Lite
Strong and inexpensive GPT-6 Astra, $20, position 95 of 100. GPT-5.5 Pro, $68, position 92 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 92 of 100. GPT-5.4 Mini, $1.7, position 65 of 100. GPT-5.6 Luna, $0.45, position 56 of 100. GPT-5 Nano, $0.14, position 35 of 100. Claude Fable 5.1, $20, position 98 of 100. Claude Opus 5, $10, position 97 of 100. Claude Sonnet 5, $4, position 86 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 83 of 100. Gemini 3.8 Flash, $1.5, position 71 of 100. Gemini 3.5 Flash, $3.4, position 63 of 100. Gemini 3.5 Flash-Lite, $0.85, position 54 of 100. Gemini 2.5 Pro, $3.4, position 74 of 100. Gemini 2.5 Flash-Lite, $0.17, position 47 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-5.5 Pro GPT-5.6 Terra GPT-5.3 Codex GPT-5.6 Luna GPT-5 Nano Claude Fable 5.1 Claude Sonnet 5 Claude Haiku 4.5 Gemini 3.5 Flash Gemini 2.5 Pro Gemini 2.5 Flash-Lite
Strong and inexpensive GPT-6 Astra, $20, position 95 of 100. GPT-5.5 Pro, $68, position 80 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 66 of 100. GPT-5.4 Mini, $1.7, position 71 of 100. GPT-5.6 Luna, $0.45, position 71 of 100. GPT-5 Nano, $0.14, position 52 of 100. Claude Fable 5.1, $20, position 94 of 100. Claude Opus 5, $10, position 90 of 100. Claude Sonnet 5, $4, position 81 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 83 of 100. Gemini 3.8 Flash, $1.5, position 80 of 100. Gemini 3.5 Flash, $3.4, position 70 of 100. Gemini 3.5 Flash-Lite, $0.85, position 63 of 100. Gemini 2.5 Pro, $3.4, position 74 of 100. Gemini 2.5 Flash-Lite, $0.17, position 58 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-6 Astra GPT-5.5 Pro GPT-5.6 Sol GPT-5.3 Codex GPT-5.6 Luna GPT-5 Nano Claude Opus 5 Claude Sonnet 5 Gemini 3.5 Flash Gemini 3.5 Flash-Lite Gemini 2.5 Pro Gemini 2.5 Flash-Lite
Strong and inexpensive GPT-6 Astra, $20, position 95 of 100. GPT-5.5 Pro, $68, position 92 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 76 of 100. GPT-5.4 Mini, $1.7, position 65 of 100. GPT-5.6 Luna, $0.45, position 60 of 100. GPT-5 Nano, $0.14, position 43 of 100. Claude Fable 5.1, $20, position 94 of 100. Claude Opus 5, $10, position 90 of 100. Claude Sonnet 5, $4, position 81 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 92 of 100. Gemini 3.8 Flash, $1.5, position 76 of 100. Gemini 3.5 Flash, $3.4, position 67 of 100. Gemini 3.5 Flash-Lite, $0.85, position 54 of 100. Gemini 2.5 Pro, $3.4, position 80 of 100. Gemini 2.5 Flash-Lite, $0.17, position 47 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-6 Astra GPT-5.6 Sol GPT-5.3 Codex GPT-5.6 Luna GPT-5 Nano Claude Sonnet 5 Claude Haiku 4.5 Gemini 3.1 Pro Gemini 3.5 Flash-Lite Gemini 2.5 Flash-Lite
Strong and inexpensive GPT-6 Astra, $20, position 98 of 100. GPT-5.5 Pro, $68, position 92 of 100. GPT-5.6 Sol, $8, position 85 of 100. GPT-5.6 Terra, $4.5, position 79 of 100. GPT-5.3 Codex, $4.8, position 84 of 100. GPT-5.4 Mini, $1.7, position 65 of 100. GPT-5.6 Luna, $0.45, position 60 of 100. GPT-5 Nano, $0.14, position 33 of 100. Claude Fable 5.1, $20, position 94 of 100. Claude Opus 5, $10, position 95 of 100. Claude Sonnet 5, $4, position 84 of 100. Claude Haiku 4.5, $2, position 68 of 100. Gemini 3.1 Pro, $4.5, position 83 of 100. Gemini 3.8 Flash, $1.5, position 71 of 100. Gemini 3.5 Flash, $3.4, position 63 of 100. Gemini 3.5 Flash-Lite, $0.85, position 54 of 100. Gemini 2.5 Pro, $3.4, position 74 of 100. Gemini 2.5 Flash-Lite, $0.17, position 41 of 100 Strong and inexpensive $0.10 $0.30 $1 $3 $10 $30 $100 0 25 50 75 100 Cost per million tokens — expensive left, cheap right Position (0–100) GPT-6 Astra GPT-5.6 Sol GPT-5.6 Terra GPT-5.6 Luna GPT-5 Nano Claude Fable 5.1 Claude Haiku 4.5 Gemini 3.5 Flash Gemini 3.5 Flash-Lite Gemini 2.5 Pro Gemini 2.5 Flash-Lite
Strong and inexpensive Gemini 3 Pro Image, $0.19, position 88 of 100. Gemini 3.1 Flash Image, $0.10, position 78 of 100. Veo 3.1 Standard, $2.5, position 92 of 100. Veo 3.1 Fast, $1, position 80 of 100. Veo 3.1 Lite, $0.33, position 66 of 100. Gemini 2.5 Flash Image, $0.04, position 70 of 100 Strong and inexpensive $0.03 $0.10 $0.30 $1 $3 0 25 50 75 100 Cost per generated output — expensive left, cheap right Position (0–100) Gemini 3 Pro Image Gemini 3.1 Flash Image Veo 3.1 Standard Veo 3.1 Fast Veo 3.1 Lite Gemini 2.5 Flash Image
General purpose Model Provider Cost Position Source GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 95 Price list Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 94 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 92 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 90 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 83 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 81 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 76 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 74 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 71 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 65 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 63 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 60 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 54 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 47 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 43 Price list
Knowledge work Model Provider Cost Position Source GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 99 Price list Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 96 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 95 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 90 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 83 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 81 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 76 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 74 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 71 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 65 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 63 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 60 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 48 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 39 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 34 Price list
Software Model Provider Cost Position Source Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 98 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 97 Price list GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 95 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 92 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 92 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 86 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 83 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 74 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 71 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 65 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 63 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 56 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 54 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 47 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 35 Price list
Customer interaction Model Provider Cost Position Source GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 95 Price list Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 94 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 90 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 83 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 81 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 80 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 80 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 74 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 71 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 71 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 70 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 66 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 63 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 58 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 52 Price list
Documents and data Model Provider Cost Position Source GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 95 Price list Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 94 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 92 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 92 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 90 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 81 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 80 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 76 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 76 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 67 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 65 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 60 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 54 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 47 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 43 Price list
Autonomous workflows Model Provider Cost Position Source GPT-6 Astra Positioned at the top on reasoning-heavy work; leads published reasoning indices. OpenAI $2010 in / 50 out 98 Price list Claude Opus 5 Leads head-to-head coding arena play. Anthropic $105 in / 25 out 95 Price list Claude Fable 5.1 Anthropic's top tier; strongest published SWE-Bench Verified result in the family. Anthropic $2010 in / 50 out 94 Price list GPT-5.5 Pro The most expensive option here by a wide margin. Justified only where a wrong answer costs more than the tokens. OpenAI $6830 in / 180 out 92 Price list GPT-5.6 Sol OpenAI $84 in / 20 out 85 Price list GPT-5.3 Codex Specialised for code; priced as a mid-tier general model. OpenAI $4.81.75 in / 14 out 84 Price list Claude Sonnet 5 Anthropic $42 in / 10 out 84 Price list Gemini 3.1 Pro Price shown is the under-200k-token band; long context costs more. Google $4.52 in / 12 out 83 Price list GPT-5.6 Terra OpenAI $4.52 in / 12 out 79 Price list Gemini 2.5 Pro Previous generation, still widely deployed. Under-200k band. Google $3.41.25 in / 10 out 74 Price list Gemini 3.8 Flash Promotional pricing until 31 December 2026. Google $1.50.75 in / 3.75 out 71 Price list Claude Haiku 4.5 Anthropic $21 in / 5 out 68 Price list GPT-5.4 Mini OpenAI $1.70.75 in / 4.5 out 65 Price list Gemini 3.5 Flash Google $3.41.5 in / 9 out 63 Price list GPT-5.6 Luna The cheapest current-generation flagship-family model on this chart. OpenAI $0.450.2 in / 1.2 out 60 Price list Gemini 3.5 Flash-Lite Google $0.850.3 in / 2.5 out 54 Price list Gemini 2.5 Flash-Lite Google $0.170.1 in / 0.4 out 41 Price list GPT-5 Nano OpenAI $0.140.05 in / 0.4 out 33 Price list
Media and content Model Provider Cost Position Source Veo 3.1 Standard $0.40-$0.60 per second. Plotted as a five-second clip at the midpoint. Google $2.5 92 Price list Gemini 3 Pro Image $0.134-$0.24 per image depending on resolution; the midpoint is plotted. Google $0.19 88 Price list Veo 3.1 Fast $0.10-$0.30 per second. Plotted as a five-second clip at the midpoint. Google $1 80 Price list Gemini 3.1 Flash Image $0.045-$0.151 per image depending on resolution; the midpoint is plotted. Google $0.10 78 Price list Gemini 2.5 Flash Image Priced per generated image. Google $0.04 70 Price list Veo 3.1 Lite $0.05-$0.08 per second. Plotted as a five-second clip at the midpoint. Google $0.33 66 Price list
Data checked on 7 September 2026.
How to read this The cost is factual. Every price comes from the provider’s own pricing page, with the date we checked it and a link to check it again. Shown is a blend of three parts input to one part output — the shape of typical drafting and retrieval work. An agentic workload inverts that ratio and shifts the cost to the left.
The position is not a benchmark score. No public benchmark scores all of these models on all of these kinds of work, and inventing a number to fill that gap would be the only dishonest thing in this tool. The position is our assessment, anchored on the capability ladder each provider publishes itself.
Image and video models charge per output, not per token. The media view puts one image beside a five-second clip; where a provider quotes a range, the midpoint is shown. Those assumptions are stated here because they decide the number.
