Pick for Coding autocomplete: DeepSeek V4.1 Flash at $8.99 per month. Best cited coding score for a short, frequent completion: Terminal-Bench 85.8 and DeepSWE 68.7 (huggingface.co). Prices from Models.dev. Data last refreshed Oct 16, 2026, 07:15 AM UTC.
Job
Which model should I use for fast coding autocomplete, and what will it cost each month?
The score uses the same cited software-engineering ladder as coding agents, then subtracts a log of the monthly cost. A short completion happens often, so a cheaper cited coder rises. Cards are chosen only from models with a cited coding score. The premium card drops the price term.
1,500 short completions a day. Each one reads about 1,200 tokens of nearby code and writes about 180. Half of the input is treated as a cached prefix.
214 models cleared the bar.304 failed the lab, provider, or specialty bar.3 failed a capability filter.1 listed at $0 were left out.0 had no usable price.
Top pick
DeepSeek V4.1 Flash$8.99 / month
Best cited coding score for a short, frequent completion: Terminal-Bench 85.8 and DeepSWE 68.7 (huggingface.co).
Score 44.9 · $0.15 input · $0.60 output per 1M · context 1,000,000 · 67 providers
Budget pick
GLM-5.3-Flash$8.91 / month
Lower bill with a cited coding result: Terminal-Bench 84.3 and DeepSWE 63.4 (huggingface.co).
Score 39.1 · $0.15 input · $0.50 output per 1M · context 1,000,000 · 74 providers
Premium pick
Claude Fable 5$702 / month
Highest cited coding ladder score if you spend more: SWE-Bench Pro 80.3 and Terminal-Bench 88 (anthropic.com).
Score 9.6 · $10.00 input · $50.00 output per 1M · context 1,000,000 · 43 providers
Skip this page for a multi-step coding agent that edits a repo. That job is coding agents, and it does not reward the cheaper bill the same way. A cited SWE-Bench score is not a measurement of keystroke latency.
The reference workload is 1,500 completions a day, about 1,200 tokens in and 180 out, with a 50% cache hit. The cards reprice when you change those inputs.
The three cards follow the inputs. Fit, and who holds each role, stays on the reference workload.
| Caps | ||||||||
|---|---|---|---|---|---|---|---|---|
| 44.9 | DeepSeek V4.1 Flash | $8.99 | $0.15 | $0.60 | 1,000,000 | 67 | Sep 10, 2026 | |
| 43.0 |  Qwen3.8 Flash Next | $10.80 | $0.20 | $0.50 | 262,144 | 5 | Aug 27, 2026 | |
| 40.9 | GPT-5.6 Luna | $15.66 | $0.20 | $1.20 | 1,050,000 | 46 | Jul 9, 2026 | |
| 39.1 | GLM-5.3-Flash | $8.91 | $0.15 | $0.50 | 1,000,000 | 74 | Aug 26, 2026 | |
| 37.1 | DeepSeek V4 Flash Vision Exp | $11.25 | $0.22 | $0.65 | 1,000,000 | 16 | Aug 21, 2026 | |
| 36.2 | DeepSeek V4 Flash 0731 | $6.12 | $0.14 | $0.28 | 1,000,000 | 55 | Jul 31, 2026 | |
| 31.2 |  Muse Spark 1.3 | $72.23 | $1.25 | $4.25 | 1,048,576 | 16 | Sep 2, 2026 | |
| 30.0 | DeepSeek V4 Pro 0813 | $34.45 | $0.66 | $1.98 | 1,000,000 | 43 | Aug 12, 2026 | |
| 25.9 | Qwen3.8 Max | $109 | $2.00 | $6.00 | 1,000,000 | 32 | Aug 3, 2026 | |
| 23.2 | Grok 4.5 | $111 | $2.00 | $6.00 | 500,000 | 25 | Jul 8, 2026 | |
| 21.2 | DeepSeek V4 Flash | $6.86 | $0.14 | $0.28 | 1,000,000 | 52 | Apr 24, 2026 | |
| 20.3 | DeepSeek Chat | $8.99 | $0.15 | $0.60 | 1,000,000 | 6 | Dec 1, 2025 |
