Anthropic · Claude Opus · Frontier
Claude Opus 4.1 pricing and limits
The previous Opus generation at a much higher rate than Opus 4.5. Still pinned in some production systems; worth re-pricing because the same workload often costs far less on the newer model.
| Line | Tokens | Rate / 1M | Cost |
|---|---|---|---|
| Input, uncached | 4,000 | $15.00 | $0.0600 |
| Input, cache read | 26,000 | $1.50 | $0.0390 |
| Output | 2,500 | $75.00 | $0.1875 |
- Input / 1M
- $15.00
- Output / 1M
- $75.00
- Cached / 1M
- $1.50
- Cache write / 1M
- $18.75
- Blended 3:1
- $30.00
- Context
- 200K
What drives Claude Opus 4.1 cost
Output is the dominant line item: Opus 4.1 output costs five times its input rate, and extended thinking is billed as output. A verbose system prompt asking for detailed reasoning on every turn is the most expensive configuration you can choose.
Cache writes cost 1.25x input for the five-minute TTL. At this input price a missed cache is expensive, so a prefix that changes per request turns caching from a saving into a surcharge.
Cost of a typical call
Take a agent step with 30,000 input tokens, 26,000 of them served from cache, and 2,500 output tokens. At Claude Opus 4.1 list prices ($15.00 input, $1.50 cached input, $75.00 output per 1M) one call costs $0.2865, against $0.6375 with no cache hits. At 20,000 calls a month that is $5,730, or $12,750 uncached.
Blended at a 3:1 input to output ratio, Claude Opus 4.1 costs $30.00 per 1M tokens. Routing the same call to the fallback in the policy below, Claude Opus 4.5, costs $0.0955, which is what a tenant pays you in COGS once they cross the hard cap and PALM SOUQ downgrades instead of blocking.
Cost comparison
Claude Opus 4.1 against three alternatives
The same agent step priced on Claude Opus 4.5, Claude Sonnet 4.5, o3. Move the sliders to match your traffic; models without a cache-read rate bill every input token in full.
Benchmark quality per task before switching. A cheaper model that needs more steps or longer outputs can cost more per finished task than its rate suggests, which is why PALM SOUQ reports cost per feature, not only per token.
Claude Opus 4.1 this guide
$0.2852 per call
$7,129 / mo
baseline
Claude Opus 4.5
$0.0950 per call
$2,376 / mo
-67% vs Claude Opus 4.1
Claude Sonnet 4.5
$0.0570 per call
$1,426 / mo
-80% vs Claude Opus 4.1
o3
$0.0408 per call
$1,021 / mo
-86% vs Claude Opus 4.1
List prices from the PALM SOUQ LLM Price Index, computed per call as uncached input + cache reads + output. Reasoning tokens count as output on reasoning models.
Rate limits by usage tier
Per-model RPM, input TPM and output TPM by usage tier. Opus models typically sit at lower ceilings than Sonnet or Haiku at the same tier, so bursty traffic is more likely to see 429s.
Exact limits vary by account tier and change over time, so we do not reprint them here. Check the limits page for your own account in the Anthropic documentation at docs.anthropic.com. PALM SOUQ enforces your own per-tenant limits in front of the provider's, so a single customer cannot consume the shared allowance that every other tenant depends on.
Circuit-breaker best practices
Apply a strict per-tenant monthly dollar cap with automatic fallback. At Opus 4.1 rates a single stuck agent session can consume a month of a mid-tier plan's margin in an afternoon.
Watch for self-reflection spirals, where the agent critiques and rewrites its own answer repeatedly. The semantic echo check spots near-identical drafts and stops the cycle.
When a breaker trips, the gateway answers in under 50 ms with HTTP 429 and X-PalmSouq-Reason: circuit_breaker_triggered, before the request reaches Anthropic. Your agent framework sees a normal rate-limit response and stops, and the audit log records which heuristic fired and for which tenant.
# palmsouq.policy.yaml
policy: claude-opus-4-1-enterprise
match:
model: claude-opus-4.1
tier: enterprise # from X-PalmSouq-Tier
budgets:
monthly_usd: 1500 # per tenant
soft_cap_pct: 80 # webhook to your app
on_hard_cap: fallback # or: block (HTTP 429)
fallback:
- claude-opus-4.5
breakers:
tool_thrashing: { identical_calls: 3, window_steps: 8 }
token_velocity: { max_tokens_per_s: 50000, window_s: 10 }
semantic_echo: { repeats: 3 }
max_steps: 25
cache:
meter_reads: true
meter_writes: true
alert_read_share_below: 0.6- budgetsPer-tenant monthly dollar cap with an 80% soft-cap webhook, keyed by X-PalmSouq-Tenant.
- fallbackAt the hard cap, requests reroute to Claude Opus 4.5 ($10.00 blended per 1M) instead of failing.
- breakersAll four loop heuristics on, evaluated in the request path with under 5 ms added latency.
Where it fits
Keep only where you have measured a quality difference that Opus 4.5 does not match. Otherwise migrate and use the saving to fund caching and evaluation.
Provider documentation: docs.anthropic.com
Claude Opus 4.1 pricing questions
How much does Claude Opus 4.1 cost per million tokens?
At list price Claude Opus 4.1 costs $15.00 per 1M input tokens and $75.00 per 1M output tokens, with cached input at $1.50 per 1M and five-minute cache writes at $18.75 per 1M. Prices were last reviewed in the PALM SOUQ LLM Price Index and are for the standard tier and shortest context band.
Does Claude Opus 4.1 support prompt caching?
Yes. Cache reads cost $1.50 per 1M tokens and cache writes cost $18.75 per 1M for the five-minute TTL. Caching saves money only when a prefix is read more often than it is rewritten, which is why PALM SOUQ meters reads and writes separately per tenant.
What are the Claude Opus 4.1 rate limits?
Per-model RPM, input TPM and output TPM by usage tier. Opus models typically sit at lower ceilings than Sonnet or Haiku at the same tier, so bursty traffic is more likely to see 429s. For your account's exact numbers, see the Anthropic documentation at docs.anthropic.com.
How do I stop runaway Claude Opus 4.1 agent loops?
Apply a strict per-tenant monthly dollar cap with automatic fallback. At Opus 4.1 rates a single stuck agent session can consume a month of a mid-tier plan's margin in an afternoon. The sample policy on this page enables all four PALM SOUQ heuristics with a max step cap of 25 and falls back to Claude Opus 4.5 at the hard cap.
Is there a batch discount for Claude Opus 4.1?
50% off input and output through the Message Batches API. Batch traffic suits evaluations, backfills and nightly enrichment jobs that do not need a real-time answer.
Govern Claude Opus 4.1 spend per customer.
Point your Anthropic client at gateway.palmsouqltd.com, pass X-PalmSouq-Tenant, and every Claude Opus 4.1 token is attributed, capped and protected by circuit breakers.
14-day free production trial · 100k requests included · no card required