PREVIOUS GROK1M CONTEXTPROMPT CACHINGCONFIGURABLE REASONING
Grok 4.3 API Pricing
Grok 4.3 remains listed on SpaceXSI's pricing page as the cheaper 1M-context Grok option: $1.25/M input, $0.20/M cached input, and $2.50/M output. SpaceXSI's newer 500K-context line, from Grok 4.5 on, lists higher per token.
Input - per 1M tokens
$1.25/M
Previous Grok 4.3 live
Output - per 1M tokens
$2.50/M
Reasoning tokens billed as output live
Cached input
$0.20/M
Prompt cache SpaceXSI row -84%
Effective - agentic blend
$0.558/M
92/8 split - 82% cache
§ 01 / TERMINAL
Grok 4.3 API cost calculator.
Live calculator pre-loaded with Grok 4.3 rates. After the Grok 4.5 launch, Grok 4.3 remains useful as a cheaper 1M-context option at $1.25/M input, $0.20/M cached input, and $2.50/M output.
$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
Grok 4.3 cost on real workloads.
AGENT
Tool-calling agent
$0.051/task
LONG CONTEXT
Research pack synthesis
$0.103/pack
CHAT
Power-user chat
$0.031/turn
RAG
Knowledge base RAG
$0.088/answer
Estimate · grok-tokenizer-estimate · ≈3.85 chars/token Auto-counts as you type
This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.
Characters 362
Words 64
Tokens (estimated) 94 tokens
Cost as input · uncached $0.00012 USD
Cost as output · uncached $0.00023 USD
Cost as cached input $0.00002 USD
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Grok 4.5 | $2.00 cache $0.30 | $6.00 | $1.04 +86% | 500K | Coding, agents, and knowledge work |
| Grok 4.3 Current | $1.25 cache $0.20 | $2.50 | $0.558 agentic 92/8 | 1M | Lower-cost Grok chat and tools |
| Grok Build 0.1 | $1.00 cache $0.20 | $2.00 | $0.476 −15% | 256K | Dedicated coding workflows |
| Grok 4.20 Multi-Agent (0309) | $1.25 cache $0.20 | $2.50 | $0.558 same | 1M | Pinned long-context agent loops |
| Claude Sonnet 5.5 | $2.00 cache $0.10 | $10.00 | $1.29 +2.3× | 1M | Claude-class coding agents |
| GPT-5.6 Sol | $4.00 cache $0.40 | $20.00 | $2.73 +4.9× · promo | 1.05M | OpenAI frontier workloads |
| DeepSeek V4 Pro | $1.32 cache $0.044 | $3.96 | $0.569 same | 1M | Low-cost reasoning alternative |
Grok 4.3 API pricing FAQ.
Grok 4.3 pricing questions, with SpaceXSI token rates separated from workload assumptions.
Q · 01 What is Grok 4.3's API price? +
SpaceXSI still lists
grok-4.3 at $1.25/M input, $0.20/M cached input, and $2.50/M output. The newer 500K-context Groks, from Grok 4.5 on, list higher, so Grok 4.3 remains the cheaper 1M-context option.Q · 02 Does Grok 4.3 support prompt caching? +
Yes. The SpaceXSI pricing page lists cached input at
$0.20/M, versus $1.25/M for fresh input. This page uses an 82% cache-hit assumption only for the effective blend tile.Q · 03 What happened to older Grok slugs? +
SpaceXSI says several older model slugs retired on
2026-05-15 and now redirect to grok-4.3. Requests through those deprecated slugs are billed at Grok 4.3 pricing.Q · 04 Is Batch API cheaper? +
Yes, by 20%. SpaceXSI's pricing page lists
grok-4.3 among the models with a 20% Batch API discount, applied to every token type: $1.00/M input, $0.16/M cached input and $2.00/M output. Batch jobs are queued and most complete within 24 hours. Since late October 2026 the newer Grok 4.5, 4.6 and 4.7 get the same 20%; only Grok Build 0.1 is left off the list. This page keeps the headline calculator on standard realtime rates.Q · 05 Are tool calls included in token pricing? +
No. SpaceXSI prices server-side tools separately, for example Web Search and X Search at
$5 / 1k calls. Token usage from those requests is still billed at the selected model rate.Q · 06 How accurate is the tokenizer estimate? +
The widget uses a GPT-class planning estimate of
4.875 characters per token. Exact SpaceXSI billing can vary with language, hidden reasoning tokens, tool use, and cached prompt boundaries.