OpenAI · GPT-4.1 · Small
GPT-4.1 mini pricing and limits
A small, fast GPT-4.1 variant that keeps the 1M context window. A common choice for high-volume extraction and classification where reasoning tokens would be wasted spend.
| Line | Tokens | Rate / 1M | Cost |
|---|---|---|---|
| Input, uncached | 8,000 | $0.40 | $0.00320 |
| Input, cache read | 4,000 | $0.10 | $0.00040 |
| Output | 600 | $1.60 | $0.00096 |
- Input / 1M
- $0.40
- Output / 1M
- $1.60
- Cached / 1M
- $0.10
- Cache write / 1M
- n/a
- Blended 3:1
- $0.70
- Context
- 1M
What drives GPT-4.1 mini cost
Like GPT-4.1, cost scales with prompt length rather than hidden reasoning. The trap is volume: pipelines that fan out one document into dozens of parallel extraction calls multiply input tokens quickly, and each call resends shared instructions.
Cached input is billed at a quarter of the input price. Batch-style workloads that share a long instruction block benefit, and the Batch API halves the remaining cost for anything that does not need a real-time answer.
Cost of a typical call
Take a extraction call with 12,000 input tokens, 4,000 of them served from cache, and 600 output tokens. At GPT-4.1 mini list prices ($0.40 input, $0.10 cached input, $1.60 output per 1M) one call costs $0.00456, against $0.00576 with no cache hits. At 800,000 calls a month that is $3,648, or $4,608 uncached.
Blended at a 3:1 input to output ratio, GPT-4.1 mini costs $0.70 per 1M tokens. Routing the same call to the fallback in the policy below, GPT-4.1 nano, costs $0.00114, which is what a tenant pays you in COGS once they cross the hard cap and PALM SOUQ downgrades instead of blocking.
Cost comparison
GPT-4.1 mini against three alternatives
The same extraction call priced on GPT-5 mini, Gemini 2.5 Flash, Claude Haiku 4.5. 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.
GPT-4.1 mini this guide
$0.00457 per call
$4,572 / mo
baseline
GPT-5 mini
$0.00331 per call
$3,309 / mo
-28% vs GPT-4.1 mini
Gemini 2.5 Flash
$0.00421 per call
$4,209 / mo
-8% vs GPT-4.1 mini
Claude Haiku 4.5
$0.0114 per call
$11,436 / mo
+150% vs GPT-4.1 mini
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
Organization and model level TPM and RPM, raised by usage tier. Smaller models generally receive higher ceilings, but fan-out jobs from a single tenant can still exhaust them within seconds.
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 OpenAI documentation at platform.openai.com/docs. 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
Cap sub-agent fan-out per request. PALM SOUQ counts concurrent in-flight requests per tenant and per feature header, so a misconfigured map step that spawns thousands of calls trips a breaker instead of a provider rate limit.
Use schema-validation retry limits: if your code retries whenever the JSON does not parse, cap retries per logical task and let the breaker return HTTP 429 after the limit.
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 OpenAI. 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: gpt-4-1-mini-pro
match:
model: gpt-4.1-mini
tier: pro # from X-PalmSouq-Tier
budgets:
monthly_usd: 120 # per tenant
soft_cap_pct: 80 # webhook to your app
on_hard_cap: fallback # or: block (HTTP 429)
fallback:
- gpt-4.1-nano
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: 10
cache:
meter_reads: true- budgetsPer-tenant monthly dollar cap with an 80% soft-cap webhook, keyed by X-PalmSouq-Tenant.
- fallbackAt the hard cap, requests reroute to GPT-4.1 nano ($0.18 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
Good for structured extraction, tagging, routing decisions and summarization at high volume, especially over long inputs that smaller context windows cannot hold.
Provider documentation: platform.openai.com/docs
GPT-4.1 mini pricing questions
How much does GPT-4.1 mini cost per million tokens?
At list price GPT-4.1 mini costs $0.40 per 1M input tokens and $1.60 per 1M output tokens, with cached input at $0.10 per 1M. Prices were last reviewed in the PALM SOUQ LLM Price Index and are for the standard tier and shortest context band.
Does GPT-4.1 mini support prompt caching?
Yes. Cached input is billed at $0.10 per 1M tokens against $0.40 for uncached input. PALM SOUQ records cached tokens per tenant from the provider's usage data, so cache savings show up in each customer's COGS.
What are the GPT-4.1 mini rate limits?
Organization and model level TPM and RPM, raised by usage tier. Smaller models generally receive higher ceilings, but fan-out jobs from a single tenant can still exhaust them within seconds. For your account's exact numbers, see the OpenAI documentation at platform.openai.com/docs.
How do I stop runaway GPT-4.1 mini agent loops?
Cap sub-agent fan-out per request. PALM SOUQ counts concurrent in-flight requests per tenant and per feature header, so a misconfigured map step that spawns thousands of calls trips a breaker instead of a provider rate limit. The sample policy on this page enables all four PALM SOUQ heuristics with a max step cap of 10 and falls back to GPT-4.1 nano at the hard cap.
Is there a batch discount for GPT-4.1 mini?
50% off input and output through the Batch API (24-hour completion window). Batch traffic suits evaluations, backfills and nightly enrichment jobs that do not need a real-time answer.
Govern GPT-4.1 mini spend per customer.
Point your OpenAI client at gateway.palmsouqltd.com, pass X-PalmSouq-Tenant, and every GPT-4.1 mini token is attributed, capped and protected by circuit breakers.
14-day free production trial · 100k requests included · no card required