What does a coding agent cost per month with an AI API?

On current list prices, GPT-6 Luna is the cheapest model we track for this workload at $24.80 per month, while the most expensive costs $2,480 — a 100× spread. Every model’s exact figure is in the table above, verified November 7, 2026.

Which AI model is cheapest for a coding agent?

GPT-6 Luna charges $0.10 input / $0.50 per 1M tokens, which works out to $0.0025 per request and $24.80 per month for this workload. Check the capability column too — the cheapest models can score lower on benchmarks, and a mid-priced model is often the better deal.

How is the monthly cost calculated?

Each request splits the 40,000 input tokens into fresh and cached tokens, multiplies each by the model’s rate, then adds 2,000 output tokens at the output rate: cost = (fresh input × input rate + cached input × cached rate + output × output rate) ÷ 1,000,000. The per-request cost is then multiplied by 10,000 requests per month.

How accurate is this estimate?

It uses each provider’s standard list prices, verified November 7, 2026, including published cached-input rates and long-context thresholds. It excludes cache-write surcharges, batch discounts, taxes and volume deals — heavy users can usually negotiate below list.

Coding agent: AI API cost per month

10,000 agent steps a month, ~40,000 input tokens each (70% read from the prompt cache) and ~2,000 output tokens.

Requests

10,000

per month

Tokens per request

40K in / 2K out

420M tokens per month

Cheapest model

$24.80

GPT-6 Luna per month

Priciest model

$2,480

GPT-6 Astra per month

What coding agent costs on every model

Model	Per request	Per month	Capability (ECI)
GPT-6 Luna$0.10 in / $0.50 out per 1M tokens	$0.0025	$24.80	156.3
Claude Haiku 5.5$0.10 in / $0.50 out per 1M tokens	$0.0025	$24.80	159.0
GPT-5.6 Luna$0.20 in / $1.20 out per 1M tokens	$0.0054	$53.60	156.3
DeepSeek V4.1 Flash$0.30 in / $1.20 out per 1M tokens	$0.0062	$61.68	155.6
Gemini 3.5 Flash-Lite$0.30 in / $2.50 out per 1M tokens	$0.0094	$94.40	145.2
Gemini 2.5 Flash$0.30 in / $2.50 out per 1M tokens	$0.0094	$94.40	140.8
Gemini 3.8 Flash$0.75 in / $3.75 out per 1M tokens	$0.02	$186.00	157.2
Gemini 3.6 Flash$0.75 in / $3.75 out per 1M tokens	$0.02	$186.00	154.2
GPT-6.1 Terra$1.00 in / $5.00 out per 1M tokens	$0.02	$234.00	—
Claude Haiku 4.5$1.00 in / $5.00 out per 1M tokens	$0.02	$248.00	142.4
DeepSeek V4 Pro$1.32 in / $3.96 out per 1M tokens	$0.02	$249.92	155.3
Grok 4.3$1.25 in / $2.50 out per 1M tokens	$0.03	$256.00	—
Gemini 2.5 Pro$1.25 in / $10.00 out per 1M tokens	$0.04	$385.00	145.3
GPT-6.1 Sol$2.00 in / $10.00 out per 1M tokens	$0.05	$468.00	166.3
Gemini 4 Argon$2.00 in / $10.00 out per 1M tokens	$0.05	$468.00	164.6
GPT-6 Sol$2.00 in / $10.00 out per 1M tokens	$0.05	$496.00	162.6
Claude Sonnet 5.5$2.00 in / $10.00 out per 1M tokens	$0.05	$496.00	164.4
Claude Sonnet 5$2.00 in / $10.00 out per 1M tokens	$0.05	$496.00	156.4
Grok 4.7$2.00 in / $6.00 out per 1M tokens	$0.05	$500.00	154.4
Grok 4.6$2.00 in / $6.00 out per 1M tokens	$0.05	$500.00	156.3
GPT-5.6 Terra$2.00 in / $12.00 out per 1M tokens	$0.05	$536.00	159.4
Gemini 3.1 Pro$2.00 in / $12.00 out per 1M tokens	$0.05	$536.00	154.7
Claude Opus 5.5$4.00 in / $20.00 out per 1M tokens	$0.09	$936.00	166.8
GPT-5.6 Sol$4.00 in / $20.00 out per 1M tokens	$0.10	$992.00	161.7
Claude Opus 5$5.00 in / $25.00 out per 1M tokens	$0.12	$1,240	162.6
Claude Fable 5.1$10.00 in / $50.00 out per 1M tokens	$0.23	$2,270	164.7
GPT-6 Astra$10.00 in / $50.00 out per 1M tokens	$0.25	$2,480	166.4

ECI = Epoch Capabilities Index (Epoch AI, CC BY 4.0). A dash means the model is not yet rated. See the full leaderboard.

How this estimate works

The workload above — 10,000 requests a month, 40,000 input tokens and 2,000 output tokens each, with 70% of input tokens served from the prompt cache— is run through every model’s official price list. Where a provider publishes a long-context threshold and this workload’s prompts exceed it, the higher long-context rate applies automatically.

Run your own numbers in the cost calculator, or see the full methodology. If two models are close in price, compare them directly: GPT-6 Luna vs Claude Haiku 5.5.

Other workload estimates

Frequently asked questions

What does a coding agent cost per month with an AI API?: On current list prices, GPT-6 Luna is the cheapest model we track for this workload at $24.80 per month, while the most expensive costs $2,480 — a 100× spread. Every model’s exact figure is in the table above, verified November 7, 2026.
Which AI model is cheapest for a coding agent?: GPT-6 Luna charges $0.10 input / $0.50 per 1M tokens, which works out to $0.0025 per request and $24.80 per month for this workload. Check the capability column too — the cheapest models can score lower on benchmarks, and a mid-priced model is often the better deal.
How is the monthly cost calculated?: Each request splits the 40,000 input tokens into fresh and cached tokens, multiplies each by the model’s rate, then adds 2,000 output tokens at the output rate: cost = (fresh input × input rate + cached input × cached rate + output × output rate) ÷ 1,000,000. The per-request cost is then multiplied by 10,000 requests per month.
How accurate is this estimate?: It uses each provider’s standard list prices, verified November 7, 2026, including published cached-input rates and long-context thresholds. It excludes cache-write surcharges, batch discounts, taxes and volume deals — heavy users can usually negotiate below list.