Head-to-head · Cost · Speed · Flexibility
Kepler vs the alternatives
Real benchmarks, real costs. See how Kepler compares to every major coding agent and IDE.
Kepler in action
Sub-agents, tool cards, approval flow, credit tracking — all in your terminal.
Benchmarks at a glance
| Agent | SWE-bench | Cost (300) | Per issue | BYOK | OSS |
|---|---|---|---|---|---|
| Kepleryou | 47.3% | $4.14 | ~14s | Yes | Yes |
| Claude Code | 72.0% | ~$450 | ~10 min | No | No |
| Codex CLI | 69.1% | ~$250 | ~5 min | No | Yes |
| Aider | 53.0% | ~$75 | ~3 min | Yes | Yes |
| SWE-agent | 23.0% | ~$100 | ~2 min | Yes | Yes |
SWE-bench Lite · Kepler v2.3 · Updated 2026-07-15
Kepler across models
Same agent. Same harness. Only the model changes.
| Metric | DeepSeek V4 Flashnow | Tencent HY3now | GLM 5.2soon |
|---|---|---|---|
| Vendor | DeepSeek | Tencent (free) | Zhipu |
| SWE-bench | 47.3% | 44.3% | — |
| Cost / 300 | $4.14 | $0 (free) | — |
| Per instance | $0.014 | $0 | — |
| Cache hit | 91% | 93% | — |
| Date | 2026-07-15 | 2026-07-15 | pending |
Every model runs the same harness · Docker eval · SWE-bench Lite 300
CLI & IDE feature comparison
| Feature | Kepler | Claude Code | Codex CLI | Cursor | Aider |
|---|---|---|---|---|---|
| Interface | CLI + Web | CLI | CLI | IDE | CLI |
| Model choice | Any (40+) | Claude only | GPT only | Any | Any |
| BYOK | ✓ | — | — | ✓ | ✓ |
| MCP servers | ✓ | ✓ | — | ✓ | — |
| Multi-file edits | ✓ | ✓ | ✓ | ✓ | ✓ |
| Auto test + fix | ✓ | ✓ | ✓ | — | Partial |
| Sub-agents | ✓ (explore/plan/code) | ✓ | — | — | — |
| Visual workflows | ✓ (Studio) | — | — | — | — |
| Scheduled runs | ✓ (cron) | — | — | — | — |
| Web dashboard | ✓ | — | — | — | — |
| Workspace agents | ✓ | — | — | — | — |
| Session memory | ✓ (cross-session) | ✓ | — | ✓ | — |
| Git integration | ✓ | ✓ | ✓ | ✓ | ✓ |
| Sandbox execution | ✓ (Docker/MicroVM) | ✓ | ✓ (cloud) | — | — |
| Self-host | ✓ | — | — | — | ✓ |
| Open source | ✓ (MIT) | — | ✓ | — | ✓ |
| Free tier | ✓ (50 cr) | — | ✓ | ✓ | ✓ (BYOK) |
| Pricing model | Credits | Subscription | Subscription | Subscription | BYOK only |
Feature comparison as of 2026-07-15. Competitive features change frequently.
Cost breakdown
The bottom line
Kepler is 100× cheaper than Claude Code for the same SWE-bench workload ($4.14 vs ~$450). You bring your own key and pick the cheapest capable model.
Kepler is the only agent with a full platform — CLI + visual workflows + scheduled automation + domain workspaces. Others are CLI-only.
Kepler is open source and self-hostable. No vendor lock-in, no per-seat pricing, no data leaving your infra.
Try Kepler
50 free credits. BYOK for unlimited.
npm i -g @axplusb/keplerWeb app