Last updated: October 20, 2026 · Prices checked against Perplexity's API docs and changelog, Black Forest Labs' pricing docs and release notes, and Oxyy's live pages on this date.
Disclosure: This guide is published by the Oxyy.ai team. Oxyy is an independent gateway, not affiliated with Perplexity or Black Forest Labs.
Quick answer: Perplexity's Sonar Chat Completions API reached end of support on September 27, 2026. Its last rate card charged per token plus a per-request search fee (Sonar: $1 / $1 per 1M tokens + $5–$12 per 1,000 requests; Sonar Pro: $3 / $15 + $6–$14). Its replacement, the Agent API, bills each model's token rate plus tools: web search is $2.50 per 1,000 calls ($1.00 with Fast Search). For images, FLUX 3 Image costs $0.048 per 1-megapixel image and FLUX 3 Image Klein costs $0.009. FLUX.2 [klein] 9B costs from $0.015 on BFL and $0.006 per megapixel on Oxyy.
| API | Price | Unit |
|---|---|---|
Perplexity Agent API: web_search | $2.50 ($1.00 Fast Search) | per 1,000 calls, plus model tokens |
| Perplexity Search API | $5.00 ($1.00 Fast Search) | per 1,000 requests |
| Sonar (last published rate card) | $1 / $1 + $5 / $8 / $12 | per 1M tokens + per 1,000 requests (low / medium / high context) |
| Sonar Pro (last published rate card) | $3 / $15 + $6 / $10 / $14 | per 1M tokens + per 1,000 requests |
| FLUX 3 Image (BFL) | $0.041 – $0.607 | per image (768sq to 4K) |
| FLUX 3 Image Klein (BFL) | $0.008 – $0.036 | per image (768sq to 4K) |
| FLUX.2 [klein] 9B | BFL from $0.015 · Oxyy $0.006 | per megapixel |
| FLUX.2 [klein] 4B | BFL from $0.014 · Oxyy $0.0056 | per megapixel |
Key Takeaways
- Sonar is being replaced. Perplexity ended Sonar Chat Completions support on September 27, 2026. Sync and streaming requests are being reformulated as Agent API requests; async Sonar requests are no longer supported.
- The Agent API is cheaper for search. Web search costs $2.50 per 1,000 calls, or $1.00 with Fast Search, plus the chosen model's token rate.
- The Search API returns raw results at $5 per 1,000 requests ($1 with Fast Search) with no token charges.
- FLUX 3 Image launched October 1, 2026 at $0.048 per 1MP image; the open-weight FLUX 3 Image Klein launched October 15 at $0.009.
- FLUX.2 bills by megapixel, and resolution always rounds up: a 1920×1080 image (2.07MP) is charged as 3MP.
- Oxyy offers FLUX.2 [klein] at 60% off: $0.006 per megapixel for the 9B model, on the same key as 170+ text, voice and image models.
Part 1: Perplexity Sonar API Pricing
What Happened to the Sonar API?
Perplexity announced the migration on August 13, 2026 with a 45-day notice period. Its docs now say: "Sonar Chat Completions support ended on September 27, 2026." Synchronous and streaming Sonar requests keep working because they're being reformulated as Agent API requests, rolling out gradually by model. Asynchronous Sonar requests are no longer supported and should move to the Agent API's background mode.
Gateway operators reported that the sonar-pro and sonar-reasoning-pro model IDs stopped being routable on September 27, while perplexity/sonar moved onto the Agent API transport on September 25. If you still call Sonar models, plan your migration now.
Sonar's Last Published Rate Card
Sonar billed three things: input tokens, output tokens, and a per-request fee that depended on the search context size (low, medium or high).
| Model | Input (per 1M) | Output (per 1M) | Request fee per 1,000 (low / medium / high) |
|---|---|---|---|
| Sonar | $1 | $1 | $5 / $8 / $12 |
| Sonar Pro | $3 | $15 | $6 / $10 / $14 |
| Sonar Reasoning Pro | $2 | $8 | $6 / $10 / $14 |
| Sonar Deep Research | $2 | $8 | Plus $2 per 1M citation tokens, $5 per 1,000 search queries and $3 per 1M reasoning tokens |
Example from Perplexity's own docs: one Sonar Pro request with 26 prompt tokens, 832 completion tokens and low search context cost $0.000078 (input) + $0.01248 (output) + $0.006 (request fee) = $0.018558. At 10,000 requests, that's about $185.58.
Perplexity Agent API Pricing (Sonar's Replacement)
The Agent API gives you one interface to Perplexity's Sonar model plus models from OpenAI, Anthropic, Google, xAI, Z.AI, Moonshot AI and NVIDIA, with built-in tools and research presets. You pay each model's token rate plus tool fees:
| Tool | Price |
|---|---|
web_search (standard) | $0.0025 per call ($2.50 per 1,000) |
web_search (Fast Search) | $0.001 per call ($1.00 per 1,000) |
image_search | $0.0025 per successful call |
fetch_url | $0.0005 per call |
people_search / finance_search | $0.005 per call |
sandbox | $0.03 per session (20-minute billing window) |
The flex service tier multiplies token rates by 0.5× and priority by 2× on models that support them. Tool fees don't change.
Which Agent API Preset Replaces Each Sonar Model?
| Sonar model | Agent API preset | Best for |
|---|---|---|
| Sonar | fast | Single-fact lookups, definitions, quick summaries |
| Sonar Pro | fast | Everyday research and light multi-step lookups |
| Sonar Reasoning Pro | low | Multi-hop browsing and wide aggregation |
| Sonar Deep Research | high | Expert-level reasoning and exhaustive sources |
For the strongest deep research, Perplexity recommends the xhigh preset.
Perplexity Search API Pricing
If you only need raw web results to feed your own model, the Search API costs $5 per 1,000 requests, or $1 with Fast Search. One request can contain up to five queries and still counts as one billing unit. Invalid requests, rate-limited requests and upstream failures aren't billed, and there are no token charges.
Search Cost Comparison: 10,000 Searches
| Option | Search cost | Plus |
|---|---|---|
| Sonar Pro (old rate card, low context) | $60 in request fees | ~$125 in tokens (Perplexity's sample request) |
Agent API web_search | $25 | Model tokens at the chosen model's rate |
| Agent API Fast Search | $10 | Model tokens |
| Search API | $50 | Nothing (raw results only) |
| Search API Fast Search | $10 | Nothing |
Sonar on Oxyy
Oxyy lists Sonar, Sonar Pro, Sonar Reasoning Pro and Sonar Deep Research in its catalog, but marks all four as deprecated, in line with Perplexity's migration. For new search features, use Perplexity's Agent API or Search API directly. If you want search-grounded answers inside the same Oxyy key, models such as Gemini 3.1 Pro (Search Grounding) and Claude Sonnet 5 (web search tool) list search as a capability; check each model page for any per-search fees.
Part 2: Black Forest Labs FLUX API Pricing
How FLUX Pricing Works
- Credits: BFL bills in credits, where 1 credit = $0.01. The price is the same in the API and the Playground.
- FLUX 3 Image and FLUX 3 Klein: flat price per image by resolution tier.
- FLUX.2: priced by output megapixels (1024×1024 = 1MP). Resolution always rounds up to the next megapixel.
- FLUX.2 [klein]: the first megapixel costs a flat rate, and each extra megapixel adds a small amount (for example, a 2MP Klein 4B image costs $0.014 + $0.001 = $0.015).
- Batch requests multiply the base cost by the number of images.
- Video: FLUX 3 Video is priced per second of output.
FLUX 3 Image Pricing (Released October 1, 2026)
FLUX 3 Image adds bounding-box layout control in the prompt, pixel-exact local edits, and up to ten reference images per request.
| Resolution | Output size | Price per image |
|---|---|---|
768sq | 768 × 768 | $0.041 |
1k | ~1 megapixel | $0.048 |
2k | ~4 megapixels | $0.100 |
4k | ~16 megapixels | $0.607 |
A 1.5k tier (~2MP) also exists; the API returns the exact cost of each request in the submission response.
FLUX 3 Image Klein Pricing (Released October 15, 2026)
FLUX 3 Image Klein is a 5.6B-parameter open-weight version of FLUX 3 Image (FLUX Kommunity License) that runs on consumer GPUs down to an RTX 3090. It supports the same features: text to image, editing, multi-reference and bounding boxes. Text to image and editing cost the same.
| 768 × 768 | 1K | 1.5K | 2K | 4K |
|---|---|---|---|---|
| $0.008 | $0.009 | $0.011 | $0.014 | $0.036 |
FLUX.2 Pricing (Megapixel-Based)
| Model | Text to image | Image editing | Best for |
|---|---|---|---|
| FLUX.2 [klein] 4B | from $0.014 | from $0.014 | Real-time, high volume |
| FLUX.2 [klein] 9B | from $0.015 | from $0.015 | Balanced quality and speed |
| FLUX.2 [pro] | from $0.03 / MP | from $0.045 / MP | Production workflows |
| FLUX.2 [flex] | from $0.05 / MP | from $0.05 / MP | Fine-grained control, typography |
| FLUX.2 [max] | from $0.07 / MP | from $0.07 / MP | Highest quality, grounding search |
| FLUX.2 [dev] | Open weights, local only (no hosted BFL API) | Local development | |
FLUX 3 Video Pricing (Per Second)
| Mode | Full render | Draft |
|---|---|---|
| Text to video / Image to video | $0.17/s HD, $0.29/s Full HD | $0.06/s |
| Video continuation | $0.43/s HD, $0.54/s Full HD | $0.12/s |
FLUX on Oxyy: 60% Off FLUX.2 [klein]
| Model (Oxyy ID) | Standard price | Oxyy price |
|---|---|---|
FLUX.2 [klein] 9B (flux-2-klein-9b) | $0.015 / MP | $0.006 / MP |
FLUX.2 [klein] 4B (flux-2-klein-4b) | $0.014 / MP | $0.0056 / MP |
| FLUX.2 [dev], FLUX.1 [schnell] | Listed; price not published yet | |
Oxyy prices from oxyy.ai/black-forest-labs, checked October 20, 2026 (current 60% launch promotion). Oxyy lists about 3.1 s latency for FLUX.2 [klein] 9B.
Image Cost Comparison: 10,000 Images at ~1 Megapixel
| Model / provider | Cost for 10,000 images |
|---|---|
| FLUX 3 Image (BFL, 1k) | $480 |
| FLUX.2 [pro] (BFL, from) | $300 |
| FLUX.2 [klein] 9B (BFL, from) | $150 |
| FLUX 3 Image Klein (BFL, 1K) | $90 |
| FLUX.2 [klein] 9B (Oxyy) | $60 |
| FLUX.2 [klein] 4B (Oxyy) | $56 |
Generate a FLUX Image Through Oxyy (OpenAI SDK)
from openai import OpenAI
import os
client = OpenAI(base_url="https://api.oxyy.ai/v1", api_key=os.environ["OXYY_API_KEY"])
img = client.images.generate(
model="flux-2-klein-9b",
prompt="A product photo of a ceramic coffee mug on a walnut desk, soft morning light",
size="1024x1024",
)
print(img.data[0].url)Why generate images through Oxyy:
- FLUX.2 [klein] at 60% off, billed per megapixel
- The same key also covers Nano Banana 2 (Gemini image), GLM-Image, CogView-4 and 170+ text and voice models
- OpenAI-compatible
/v1/images/generations, so existing image code works - No subscription, no minimum, credits never expire, and a free starting balance
Which Should You Use?
| If you need… | Use |
|---|---|
| Cited, search-grounded answers | Perplexity Agent API (fast preset replaces Sonar and Sonar Pro) |
| Raw web results for your own RAG pipeline | Perplexity Search API ($1–$5 per 1,000 requests) |
| Best FLUX image quality and layout control | FLUX 3 Image |
| Open weights you can self-host or fine-tune | FLUX 3 Image Klein or FLUX.2 [klein] |
| The cheapest hosted FLUX images | FLUX.2 [klein] on Oxyy ($0.0056–$0.006 per MP) |
Final Verdict
If you're budgeting for Perplexity Sonar, budget for the Agent API instead: Sonar Chat Completions support has ended, and the Agent API's $1–$2.50 per 1,000 searches is cheaper than Sonar's old $5–$14 request fees. For images, FLUX 3 Image is BFL's new quality leader at $0.048 per 1MP image, while FLUX.2 [klein] is the volume workhorse. Through Oxyy it costs $0.006 per megapixel, so 10,000 images cost about $60.
Generate FLUX images at 60% off with a free Oxyy balance →
Related: FLUX API pricing on Oxyy · FLUX.2 [klein] 9B · Sonar on Oxyy · Gemini API pricing
Frequently Asked Questions
How much does the Perplexity Sonar API cost?
Sonar's last published rate card charged $1 per 1M input and output tokens plus $5, $8 or $12 per 1,000 requests depending on search context size. Sonar Pro charged $3 / $15 per 1M tokens plus $6, $10 or $14 per 1,000 requests.
Is the Perplexity Sonar API being discontinued?
Perplexity ended Sonar Chat Completions support on September 27, 2026. Synchronous and streaming requests are being reformulated as Agent API requests, and asynchronous Sonar requests are no longer supported.
What replaced Sonar Pro?
Perplexity maps Sonar and Sonar Pro to the Agent API's fast preset, Sonar Reasoning Pro to the low preset, and Sonar Deep Research to the high preset.
How much does Perplexity web search cost in the Agent API?
The web_search tool costs $0.0025 per call ($2.50 per 1,000), or $0.001 per call ($1.00 per 1,000) with Fast Search, plus the chosen model's token cost.
How much does the Perplexity Search API cost?
The Search API costs $5 per 1,000 successful requests, or $1 per 1,000 with Fast Search. Each request can include up to five queries, and there are no token charges.
How much does the FLUX API cost?
On Black Forest Labs, FLUX 3 Image costs $0.041 to $0.607 per image, FLUX 3 Image Klein costs $0.008 to $0.036, and FLUX.2 models start at $0.014 per image for Klein 4B. On Oxyy, FLUX.2 [klein] 9B costs $0.006 per megapixel.
How is FLUX.2 priced per megapixel?
FLUX.2 prices scale with output resolution, where 1024×1024 is 1 megapixel. Resolution rounds up, so a 1920×1080 image (2.07MP) is charged as 3MP. Klein models charge a flat rate for the first megapixel plus a small amount for each extra one.
What is the cheapest FLUX model?
On BFL, FLUX 3 Image Klein at $0.008 per 768×768 image and FLUX.2 [klein] 4B from $0.014 are the cheapest. On Oxyy, FLUX.2 [klein] 4B costs $0.0056 per megapixel.
Is FLUX 3 open source?
FLUX 3 Image Klein is open-weight under the FLUX Kommunity License and runs on consumer GPUs. The full FLUX 3 Image model is available through BFL's API.
Can I use FLUX with the OpenAI SDK?
Yes, through Oxyy. Set the base URL to https://api.oxyy.ai/v1, use your Oxyy API key and call the images endpoint with a model such as flux-2-klein-9b.