Last updated: November 11, 2026 | Pricing verified: November 11, 2026
Newer option (Nov 2026): OpenAI deprecated gpt-5.3-codex on October 1, 2026, and it leaves the API on April 1, 2027; the replacement is GPT-6 Sol at $2/M input and $10/M output. Devstral 2 is no longer on Mistral's API pricing page and now powers Mistral Vibe, Mistral's CLI/IDE coding tool.
For coding agents, pricing is not just input plus output. You also pay for retries, tool calls, file context, failed patches, and long explanations. That is why the cheapest coding model is not always the best buy.
In November 2026, the picture has changed. GPT-5.3 Codex is deprecated, and OpenAI points users to GPT-6 Sol at $2/M input and $10/M output. Codestral is still listed at $0.30/M input and $0.90/M output. Devstral 2 is not on Mistral's API pricing page; you use it through Mistral Vibe instead. Mistral's Pro plan is $14.99/month.
Pricing in INR
At $1 = ₹93.96:
| Model | Input | Output | Approx input INR | Approx output INR |
|---|---|---|---|---|
| Codestral | $0.30/M | $0.90/M | ₹28.19 | ₹84.56 |
| Devstral 2 | Not listed | Not listed | - | - |
| GPT-6 Sol | $2.00/M | $10.00/M | ₹187.92 | ₹939.60 |
| GPT-5.3 Codex | Deprecated | Deprecated | - | - |
Mistral's Pro plan at $14.99/month is about ₹1,408.46. Check Mistral's site for which plan includes Vibe and its usage limits.
OpenAI output is still the painful part. A coding agent that explains every change can spend more on narration than code.
Category Winners
| Category | Winner |
|---|---|
| Cheapest coding API model | Codestral |
| Best budget code completion | Codestral |
| Best Mistral agent model | Devstral 2 (via Mistral Vibe) |
| Best OpenAI ecosystem fit | GPT-6 Sol (replaces GPT-5.3 Codex) |
| Best Indian startup setup | Codestral first, GPT-6 Sol fallback |
Who Should Use What?
Students
Use Codestral if you are experimenting with code generation through an API. Try Devstral through Mistral Vibe if you want a CLI or IDE tool. Do not start with GPT-6 Sol unless you have free credits or a clear project budget.
Developers
Use GPT-6 Sol when the model has to understand a messy repo, run tool-heavy loops, or produce reliable patches. OpenAI built it for coding and agentic work. Use Codestral for autocomplete-style work and smaller functions. If you still call gpt-5.3-codex, plan the switch before April 1, 2027.
Startups
Use a two-step system:
- Codestral for routine edits
- GPT-6 Sol for escalations
- output caps on every request
- no verbose explanations unless the user asks
That last point saves real money.
Hidden Limitation
Most coding comparisons ignore output verbosity. A model that returns a 200-line explanation after a 10-line change can be expensive in production. Force short diffs, JSON summaries, or patch-only responses when possible.
Pros and Cons
Mistral Devstral/Codestral
Pros
- Much cheaper than OpenAI's coding models
- Good fit for repetitive coding tasks
- Codestral has simple API pricing
Cons
- Devstral 2 has no listed API price; it lives inside Mistral Vibe
- May need more retry logic
- Payment flow can be less India-friendly
GPT-5.3 Codex / GPT-6 Sol
Pros
- Strong OpenAI developer tooling
- GPT-6 Sol is built for coding and agentic work
- Batch requests are 50% off
Cons
- GPT-5.3 Codex is deprecated and will be removed
- Expensive output tokens
- Not ideal for student-budget API experiments
- USD billing can be annoying from India
Final Verdict
For Indian developers, GPT-6 Sol is now the premium coding fallback, not the default worker. It replaces GPT-5.3 Codex. Use Codestral for cheap repeated work, or Devstral through Mistral Vibe if you prefer a tool over an API. Send only hard repo-level tasks to GPT-6 Sol.
Sources & Method
Pricing was verified on November 11, 2026 from the official OpenAI and Mistral pricing pages, plus OpenAI's deprecations page. INR uses ₹93.96 per USD (open.er-api.com rate, November 11, 2026). We compared cost structure for coding-agent workflows, not fake benchmark rankings.
FAQ
Is GPT-5.3 Codex expensive?
It is deprecated now. OpenAI deprecated gpt-5.3-codex on October 1, 2026 and will remove it on April 1, 2027. Its replacement, GPT-6 Sol, costs $2/M input and $10/M output, roughly ₹187.92/M input and ₹939.60/M output.
Is Devstral 2 cheaper than Codex?
Devstral 2 is no longer on Mistral's API pricing page, so there is no per-token price to compare. It powers Mistral Vibe. Mistral's Pro plan is $14.99/month; check which plan includes Vibe.
Which is best for Indian students?
Codestral. At $0.30/M input and $0.90/M output, it is much cheaper for learning and experiments.
Which is best for production coding agents?
Use a router. Start with Codestral, escalate hard tasks to GPT-6 Sol.