Cost-performance analysis

Which Model Is Best for Coding?

A cost-performance leaderboard of the models coding agents run on - benchmark scores, real token-mix pricing, latency, and decoding speed.

Models used by coding agents

Sort by:

Model	Modality	DF Coding Score	Avg Coding Task Cost
Claude Opus 5.5 Anthropic	1M	97.6	$1.19 USD / task
Claude Sonnet 5.5 Anthropic	1M	94.3	—
Claude Fable 5.1 Anthropic	1M	88.6	$2.73 USD / task
GPT-6 Astra OpenAI	1M	86.3	$1.15 USD / task
GPT-6.1 Sol OpenAI	1.05M	83.9	$0.181 USD / task
GPT-6 Sol OpenAI	872k	79.7	$0.201 USD / task
Claude Fable 5 Anthropic	1M	79.4	$2.36 USD / task
Claude Opus 5 Anthropic	1M	78.0	$0.821 USD / task
GPT-5.6 Sol OpenAI	1.05M	77.8	$0.350 USD / task
GPT-5.6 Terra OpenAI	1.05M	72.2	$0.234 USD / task
Gemini 3.8 Flash Google	1M	66.9	$0.135 USD / task
GPT-5.4 OpenAI	1M	63.3	$0.254 USD / task
GPT-5.5 OpenAI	1.05M	61.9	$0.490 USD / task
GPT-6 Luna OpenAI	1.05M	60.9	$0.00843 USD / task
Claude Opus 4.7 Anthropic	1M	60.7	$1.58 USD / task
Claude Opus 4.8 Anthropic	1M	60.4	$1.16 USD / task
GPT-5.6 Luna OpenAI	1.05M	59.6	$0.0301 USD / task
Claude Sonnet 5 Anthropic	1M	56.4	$0.969 USD / task
Gemini 3.1 Pro (Preview) Google	1M	52.1	$0.783 USD / task
Claude Opus 4.6 Anthropic	1M	51.6	$0.581 USD / task
Claude Opus 4.5 Anthropic	200k	48.1	—
Gemini 3.5 Flash Google	1M	47.8	$0.620 USD / task
Gemini 3 Flash (Preview) Google	1M	46.3	—
Claude Sonnet 4.6 Anthropic	1M	41.1	—
GPT-5.4 mini OpenAI	400k	39.3	$0.183 USD / task
Kimi-K2.7-Code Moonshot AI	256k	38.0	—
GPT-5.4 nano OpenAI	400k	28.6	—
Claude Sonnet 4.5 Anthropic	1M	27.1	—
Claude Haiku 4.5 Anthropic	200k	21.6	$0.159 USD / task
Gemini 2.5 Pro Google	1M	20.2	—
GPT-5 mini OpenAI	400k	19.2	—
GPT-OSS 120B OpenAI	131k	0.0	—
Raptor mini GitHub	400k	—	—
MAI-Code-1-Flash Microsoft		—	—
Claude Opus 4.8 Fast Anthropic	1M	—	—

Claude Opus 5.5 Anthropic
#1

DF Score 97.6

Avg task cost $1.19

Latency —

Context window 1M
Claude Sonnet 5.5 Anthropic
#2

DF Score 94.3

Avg task cost —

Latency 454.56s

Context window 1M
Claude Fable 5.1 Anthropic
#3

DF Score 88.6

Avg task cost $2.73

Latency 210.79s

Context window 1M
GPT-6 Astra OpenAI
#4

DF Score 86.3

Avg task cost $1.15

Latency —

Context window 1M
GPT-6.1 Sol OpenAI
#5

DF Score 83.9

Avg task cost $0.181

Latency —

Context window 1.05M
GPT-6 Sol OpenAI
#6

DF Score 79.7

Avg task cost $0.201

Latency 136.12s

Context window 872k
Claude Fable 5 Anthropic
#7

DF Score 79.4

Avg task cost $2.36

Latency 239.01s

Context window 1M
Claude Opus 5 Anthropic
#8

DF Score 78.0

Avg task cost $0.821

Latency 56.78s

Context window 1M
GPT-5.6 Sol OpenAI
#9

DF Score 77.8

Avg task cost $0.350

Latency —

Context window 1.05M
GPT-5.6 Terra OpenAI
#10

DF Score 72.2

Avg task cost $0.234

Latency —

Context window 1.05M
Gemini 3.8 Flash Google
#11

DF Score 66.9

Avg task cost $0.135

Latency 14.66s

Context window 1M
GPT-5.4 OpenAI
#12

DF Score 63.3

Avg task cost $0.254

Latency 113.80s

Context window 1M
GPT-5.5 OpenAI
#13

DF Score 61.9

Avg task cost $0.490

Latency 118.46s

Context window 1.05M
GPT-6 Luna OpenAI
#14

DF Score 60.9

Avg task cost $0.00843

Latency 93.68s

Context window 1.05M
Claude Opus 4.7 Anthropic
#15

DF Score 60.7

Avg task cost $1.58

Latency —

Context window 1M
Claude Opus 4.8 Anthropic
#16

DF Score 60.4

Avg task cost $1.16

Latency 30.05s

Context window 1M
GPT-5.6 Luna OpenAI
#17

DF Score 59.6

Avg task cost $0.0301

Latency —

Context window 1.05M
Claude Sonnet 5 Anthropic
#18

DF Score 56.4

Avg task cost $0.969

Latency 144.37s

Context window 1M
Gemini 3.1 Pro (Preview) Google
#19

DF Score 52.1

Avg task cost $0.783

Latency 25.74s

Context window 1M
Claude Opus 4.6 Anthropic
#20

DF Score 51.6

Avg task cost $0.581

Latency —

Context window 1M
Claude Opus 4.5 Anthropic
#21

DF Score 48.1

Avg task cost —

Latency 15.04s

Context window 200k
Gemini 3.5 Flash Google
#22

DF Score 47.8

Avg task cost $0.620

Latency 20.59s

Context window 1M
Gemini 3 Flash (Preview) Google
#23

DF Score 46.3

Avg task cost —

Latency 7.60s

Context window 1M
Claude Sonnet 4.6 Anthropic
#24

DF Score 41.1

Avg task cost —

Latency 1.16s

Context window 1M
GPT-5.4 mini OpenAI
#25

DF Score 39.3

Avg task cost $0.183

Latency 11.87s

Context window 400k
Kimi-K2.7-Code Moonshot AI
#26

DF Score 38.0

Avg task cost —

Latency 2.92s

Context window 256k
GPT-5.4 nano OpenAI
#27

DF Score 28.6

Avg task cost —

Latency 4.74s

Context window 400k
Claude Sonnet 4.5 Anthropic
#28

DF Score 27.1

Avg task cost —

Latency 1.36s

Context window 1M
Claude Haiku 4.5 Anthropic
#29

DF Score 21.6

Avg task cost $0.159

Latency 0.81s

Context window 200k
Gemini 2.5 Pro Google
#30

DF Score 20.2

Avg task cost —

Latency 22.54s

Context window 1M
GPT-5 mini OpenAI
#31

DF Score 19.2

Avg task cost —

Latency 85.30s

Context window 400k
GPT-OSS 120B OpenAI
#32

DF Score 0.0

Avg task cost —

Latency 0.94s

Context window 131k
Raptor mini GitHub
#33

DF Score —

Avg task cost —

Latency —

Context window 400k
MAI-Code-1-Flash Microsoft
#34

DF Score —

Avg task cost —

Latency —

Context window
Claude Opus 4.8 Fast Anthropic
#35

DF Score —

Avg task cost —

Latency —

Context window 1M

Help us to detect updates in plans values

The prices and token mixes above are driven by real coding usage. Share yours via letmecode to allow us discover updates in plans values.

$ npx letmecode@latest

FAQ

The agent decision

Your model choice sets the budget. The agent sets the workflow.

Compare leading coding agents on the models they bundle, limits, and pricing to see which one turns your model choice into the best engineering results.

See the agent rankings Limits, models & cost

Something unclear or missing?

If any of the numbers or terms above don't add up, or you spotted something that looks off, tell us - we'll clarify and keep the data sharper for everyone.

Report inconsistency

Found it useful - share

Share this comparison with the world to give everyone the opportunity to make the right choice based on real numbers instead of marketing claims.