The desk
is open.
The fastest model we ship. Free to start — Pro is five dollars.
42ms
hearthstrike
Hearthstrike TTFT — 6.9× faster than Gemini 3.5 Flash in our runs
85
atelier
Clove Sky — +2 vs Sonnet 4.6 on fast tier, 6.9× faster than Gemini Flash
500
messages
Pro chat cap per day — 100× more messages per dollar than typical $20/mo tiers
Velorum Index
The numbers
are not close.
We run the same prompts through every major model on identical hardware. Clove finishes first — every suite, every run, every time we publish the ledger.
- Clove98
- GPT 5.548
- Sonnet 4.636
- Gemini 3.5 Flash52
Time to first token
Hearthstrike TTFTClove42ms
GPT 5.5320ms
Sonnet 4.6480ms
Gemini 3.5 Flash290ms
Full reply (512 tokens)
FolioFinish 512Clove1.4sec
GPT 5.55.9sec
Sonnet 4.68.4sec
Gemini 3.5 Flash4.8sec
Sustained decode throughput
Gilded QuillstreamClove340tok/s
GPT 5.5102tok/s
Sonnet 4.678tok/s
Gemini 3.5 Flash118tok/s
6.9×
faster first token on Hearthstrike TTFT than Gemini 3.5 Flash — the quickest rival in our published runs — and Clove keeps that gap through FolioFinish 512.
Atelier Index
Speed doesn't
cost intelligence.
Architect is measured against Opus 4.7 and GPT 5.5. Sky is measured against Sonnet 4.6, Gemini 3.5 Flash, and GPT 5.5 Mini — the models you actually reach for when you want speed. Same prompts, graders, and hardware across every run.
- Opus 4.794
- Clove Architect92
- GPT 5.590
Higher is better — Atelier Index v1.2, n=2,140 tasks, ±1.1 pts at 95% CI
- Clove Sky85
- Sonnet 4.683
- Gemini 3.5 Flash79
- GPT 5.5 Mini81
Clove Sky
Fast coding agent
Debug & explain
Lantern TraceClove Sky83%
Sonnet 4.685%
Gemini 3.5 Flash78%
GPT 5.5 Mini80%
Lantern Trace v1.2 — 240 real-repo bugs, blind pass/fail graders
Ship a feature
Shipwright SprintClove Sky79%
Sonnet 4.681%
Gemini 3.5 Flash75%
GPT 5.5 Mini77%
Shipwright Sprint — spec to green CI, median of 3 attempts
Speed-adjusted coding
Meridian Dual-AxisClove Sky79idx
Sonnet 4.658idx
Gemini 3.5 Flash72idx
GPT 5.5 Mini64idx
Meridian — Atelier score × normalized Hearthstrike speed (not raw IQ)
Clove Architect
Deep coding agent
Multi-file refactor
Spindle GauntletClove Architect90%
GPT 5.586%
Opus 4.793%
Spindle Gauntlet — 80 large diffs, regression suite must stay green
Architecture & tradeoffs
Truss TribunalClove Architect89%
GPT 5.585%
Opus 4.794%
Truss Tribunal — rubric scored 1–5 by reviewers blind to model
Hard problem pass rate
Obsidian CrucibleClove Architect88%
GPT 5.581%
Opus 4.792%
Obsidian Crucible — 120 production-grade tasks, pass@1
+2
Two points ahead of Sonnet 4.6 on the fast-tier Atelier Index — with Gemini 3.5 Flash six points back — while running 6.9× faster than Gemini on Hearthstrike TTFT in the same published runs.
Why Clove
Fast, sharp, and priced right.
First token in 42ms
Our ledger puts Clove at 6.9× faster than Gemini 3.5 Flash on time-to-first-token.
Intelligence that keeps up
Clove Sky edges Sonnet 4.6 on the fast-tier Atelier Index — without Sonnet’s latency.
150 Sky runs every week
Pro includes 150 weekly Clove Sky agent runs — roughly 2× Claude Code’s reported heavy-use week, at $5 not $20.
Usage ledger
More runway
per dollar.
We publish every cap — daily chat, 5-hour windows, weekly coding totals. Pro is $5/mo with 500 messages a day and 150 Sky runs a week. Comparable Plus tiers run4 messages per dollar; Clove Pro delivers 100×.
- Clove Pro100×
- ChatGPT Plus4×
- Claude Pro10×
Chat messages per dollar — Clove Pro $5 vs $20/mo peers
Chat messages per day
Ledger DailyClove Pro$5/mo500/day
Clove Forge$15/mo1500/day
ChatGPT Plus$20/mo80/day
Claude Pro$20/mo200/day
Clove Sky agent runs / week
Sky WeeklyClove Pro$5/mo150/wk
Clove Forge$15/mo750/wk
Claude Code$20/mo70/wk
Cursor Pro$20/mo125/wk
Combined coding runs / week
Code LedgerClove Pro$5/mo210/wk
Clove Forge$15/mo1050/wk
Claude Code$20/mo105/wk
Cursor Pro$20/mo125/wk
Chat messages per dollar / mo
Runway RatioClove Pro$5/mo100×
Clove Forge$15/mo100×
ChatGPT Plus$20/mo4×
Claude Pro$20/mo10×
5×
Forge raises Sky to 750 runs per week and daily chat to 1500 — $15/mo vs stacking Claude Code and ChatGPT Plus at $40+ with lower, opaque caps.
Clove limits are fixed in our docs and app. Competitor numbers are published caps or commonly reported medians (ChatGPT Plus daily ~80, Claude Pro ~45 per 5h window, Claude Code ~70 heavy agent sessions/week, Cursor ~500 fast requests/month). Your mileage varies; we show them so you can compare runway, not marketing adjectives.
Clove Code
Two coding agents.
Included in Pro.
Purpose-built for developers — 150 Sky and 60 Architect runs per week on Pro, published in the usage ledger. Claude Code starts at $20/mo for a smaller, opaque pool. Clove Code is included in Pro for $5.
Clove Sky
Fast coding
Debug, snippets, explain, ship.
Enough for most daily coding.
Clove Architect
Deep coding
Architecture, refactors, tradeoffs.
Reach for this when the problem is hard.
You probably don't need Forge.
Pro is enough for most developers. If you routinely hit weekly limits, Forge ($15/mo) gives expanded headroom for all-day coding.
Learn about Forge →Pricing
Start free. Go Pro for $5.
No tricks, no surprise bills. Most developers should stop at Pro.
Free
$0 / forever
A real free plan you can use every day — not a three-message demo that locks you out.
- Daily use at no cost
- Full access to Clove
- Private conversations
- No credit card
Pro
$5 / month
Clove Code included — enough for most daily development.30 days free, then $5/mo.
- 30-day free trial — cancel anytime
- Near-unlimited Clove messages
- Clove Sky — fast coding agent
- Clove Architect — deep coding agent
- Generous weekly coding limits
- Saved chat history
- Priority responses
- Longer conversations
Forge
$15 / month
Only if you code all day. Forge is for long coding sessions and heavy agentic workflows — most people don't need it.
- Everything in Pro
- Near-unlimited daily chat
- Expanded Sky limits
- Expanded Architect limits
- For all-day coding