- Rank
- #36
- Total Output Tokens
- 523,482
- Response Time (avg)
- 32.55s
- Total Cost
- $0.355
DeepSeek V4.1 Flash 1017 (low) vs GPT-5.3-Codex (medium)
The average score is effectively tied at 9.2 vs 9.1. DeepSeek V4.1 Flash 1017 (low) has the lower benchmark cost at $0.355 vs $1.318. GPT-5.3-Codex (medium) is faster at 18.87s vs 32.55s, with pass rates of 78.3% vs 84.1%.
Compared models
- Rank
- #38
- Total Output Tokens
- 66,287
- Response Time (avg)
- 18.87s
- Total Cost
- $1.318
Recommended model
DeepSeek V4.1 Flash 1017 (low)
It has the best score here (9.2), while costing about 3.7x less than GPT-5.3-Codex (medium).
Detailed comparison
| Metric | DeepSeek V4.1 Flash 1017 DeepSeek V4.1 Flash 1017 low | GPT-5.3-Codex GPT-5.3-Codex medium |
|---|---|---|
| Score | 9.2 | 9.1 |
| Rank | #36 | #38 |
| Reliability | 10.0 | 10.0 |
| Consistency | 9.0 | 8.6 |
| Attempts | 69/69 | 69/69 |
| Tests Correct | ||
| Attempt pass rate | 78.3% | 84.1% |
| Flaky tests | 3 | 4 |
| Total Runs | 69 | 69 |
| Cost per result | 2.214 | 7.753 |
| Total Cost | $0.355 | $1.318 |
| Input Price | $0.150 / 1M | $1.750 / 1M |
| Output Price | $0.600 / 1M | $14.000 / 1M |
| Cache Read Price | $0.003 / 1M | $0.175 / 1M |
| Cache Write Price | N/A | N/A |
| Total Input Tokens | 267,122 | 222,794 |
| Output Tokens | 7,369 | 7,357 |
| Reasoning Tokens | 516,113 | 58,930 |
| Response Time (avg) | 32.55s | 18.87s |
| Response Time (max) | 443.63s | 100.93s |
| Response Time (total) | 748.62s | 433.94s |
| Parameters | ~748B total (~16B active) | ~1.2T total (~80B active) |
| Availability | Open source | Closed |
Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.
Model generation showcase
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
#36 DeepSeek V4.1 Flash 1017
low- Cost
- $0.006
- Time
- 63.8s
- Tokens
- 11,236 tok
#38 GPT-5.3-Codex
medium- Cost
- $0.049
- Time
- 54.9s
- Tokens
- 3,580 tok
Category Breakdown
| Agentic | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 46.67s | 156,869 | 2,102 | 5,290 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 61.91s | 141,607 | 1,099 | 4,814 |
| Anti-AI Tricks | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 8.3 | 10.0 | 75.0% | 0 | 2.02s | 852 | 171 | 2,193 | |
| GPT-5.3-Codex | 8.7 | 7.9 | 91.7% | 1 | 4.16s | 606 | 240 | 1,722 |
| Coding | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 27.01s | 7,509 | 323 | 45,628 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 19.50s | 7,302 | 535 | 10,890 |
| Combined | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 18.73s | 82,444 | 3,880 | 15,379 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 26.04s | 57,988 | 4,258 | 5,816 |
| Data parsing and extraction | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 1.71s | 7,446 | 207 | 1,344 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 3.07s | 7,140 | 234 | 728 |
| Domain specific | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 3.6 | 7.2 | 22.2% | 1 | 174.63s | 909 | 9 | 428,331 | |
| GPT-5.3-Codex | 5.9 | 7.2 | 55.6% | 1 | 63.93s | 732 | 71 | 30,150 |
| General Intelligence | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 2.25s | 549 | 116 | 553 | |
| GPT-5.3-Codex | 4.6 | 10.0 | 0.0% | 0 | 4.87s | 477 | 187 | 331 |
| Instructions following | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 1.25s | 783 | 69 | 898 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 3.04s | 660 | 93 | 693 |
| Puzzle Solving | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 6.9 | 4.9 | 77.8% | 2 | 2.54s | 828 | 236 | 3,620 | |
| GPT-5.3-Codex | 9.0 | 7.9 | 88.9% | 1 | 5.05s | 642 | 356 | 1,593 |
| Tool Calling | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 10.0 | 10.0 | 100.0% | 0 | 1.92s | 8,672 | 237 | 338 | |
| GPT-5.3-Codex | 10.0 | 10.0 | 100.0% | 0 | 6.37s | 5,445 | 254 | 492 |
| Trivia | Score | Consistency | Attempt pass rate | Flaky tests | Tests Correct | Response Time (avg) | Input Tokens | Output Tokens | Reasoning Tokens |
|---|---|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | 3.0 | 10.0 | 0.0% | 0 | 33.79s | 261 | 19 | 12,539 | |
| GPT-5.3-Codex | 2.8 | 1.6 | 33.3% | 1 | 14.43s | 195 | 30 | 1,701 |
Quick Compare
Switch Comparison Pair
Claude Haiku 5.5xhighvsGPT-5.3-CodexmediumGPT-5.3-CodexmediumvsQwen3.8 Max (0902)highMuse Spark 1.3highvsGPT-5.3-CodexmediumDeepSeek V4.1 Flash 1017lowvsQwen3.8 MAXmediumClaude Sonnet 5.5maxvsDeepSeek V4.1 Flash 1017lowDeepSeek V4.1 Flash 1017lowvsGPT-5.5mediumMuse WonderlowvsGPT-5.3-CodexmediumGPT-5.3-CodexmediumvsParetodefaultSeed 2.1 TurbohighvsGPT-5.3-CodexmediumDeepSeek V4.1 Flash 1017lowvsQwen3.8 Max (0902)highDeepSeek V4.1 Flash 1017lowvsMuse WondermediumClaude Opus 5.5mediumvsDeepSeek V4.1 Flash 1017low